Applying Knime Data Mining Tool to Analyse the Job and Self-satisfaction Data
Kiran Poudel
Bachelor’s Thesis Degree Programme in Business Information Technology 2021
Abstract
10 May 2021
Authors Kiran Poudel
Group BITE16S Thesis title
Applying Knime Data Mining Tool to Analyse the Job and Self- satisfaction Data
Number of pages and appendices 58
Supervisors Dr. Amir Dirin
With over four hundred thousand foreign workforces in Finland, the concerns for job as well as self-satisfaction has been growing ever since. Job and self-satisfaction have been studied for a long time. Various findings were obtained from these studies. However, the tool to analyse these data has been different and been evolved from time to time. From statistical to data mining where various tools are available for data analysis which not only display various hidden patterns inside the data but also with help of the neural engine and AI is able to predict several outcomes based on data itself. Knime being such a tool is used for data analysis for this thesis.
Results of the study find out various factors for self and job satisfaction as well as how different elements such as gender, country of birth, academic qualification and employment status plays a vital role in various life and job-related factors such as social isolation, failure, entitlement and emotions. The result also establishes a connection between these factors and figure out how they are correlated and up to what extent.
With the number of workforces predicted to grow in coming future also, the discussions about job and self-satisfaction will stay. Thus, the research aims to be useful for further study as well as various policy makings where they can relate the importance of various factors and their correlations with each other to establish a better job as well as self- satisfaction for the workforce.
Keywords
Knime, data, data mining, social isolation, failure, entitlement, emotions
Table of Contents
1 Introduction... 6
2 Research Questions and Research Methods ... 7
3 Related Research ... 9
3.1 CRISP-DM ... 10
3.2 Data Mining With Knime ... 12
3.3 Young Schema Questionnaire ... 12
3.4 Statistical vs Data Mining ... 13
4 Empirical Study/Designing the Research ... 14
5 Data Analysis ... 15
5.1 Data Preparations ... 15
6 Results ... 22
6.1 Participants Profile ... 22
6.2 Comparisons ... 28
6.3 Findings ... 36
6.4 Correlation ... 47
7 Discussions ... 49
7.1 Research Question’s Answers ... 49
7.2 Arguments ... 51
7.3 Reliability and Validity ... 52
8 Conclusions ... 54
9 References ... 55
Table of figures
Figure 1: CRISP-DM Methodology (Vorhies, 2016) ... 10
Figure 2: Excel Reader Node and its Configuration ... 15
Figure 3: Row Filter ... 16
Figure 4: Column Filter ... 17
Figure 5: Rule Engine Condition ... 18
Figure 6: Addition of New Column in Rule Engine ... 18
Figure 7: Pivoting ... 19
Figure 8: String to Number ... 20
Figure 9: Math Formula ... 20
Figure 10: Linear Correlation ... 21
Figure 11: Pie Chart (Country) ... 23
Figure 12: Pie Chart (Continent) ... 23
Figure 13: Pie Chart (Continent in Percentage) ... 24
Figure 14: Pie Chart (Gender) ... 25
Figure 15: Bar Graph (Qualification)... 25
Figure 16: Pie Chart (Qualification Rank) ... 26
Figure 17: Bar Graph (Employment Status) ... 27
Figure 18: Pie Chart (Employed or Not) ... 28
Figure 19: Bar Graph (Gender and Qualification Level) ... 29
Figure 20: Bar Graph (Gender and Employed/Not) ... 30
Figure 21: Bar Graph (Continent and Qualification Level) ... 31
Figure 22: Pie Chart (Percentage of Graduate and Under-graduate in different Continents) ... 32
Figure 23: Bar Graph (Continent and Employed/Not) ... 33
Figure 24: Pie Chart (Percentage of Employed in Different Continent) ... 34
Figure 25: Pie Chart (Percentage of Unemployed in Different Continent) ... 35
Figure 26: Bar Graph (Qualification and Employed/Not) ... 36
Figure 27: Answers in string form ... 38
Figure 28: Answers in numeric form ... 38
Figure 29: Average score of each user for different factors ... 39
Figure 30: Mean score of all users ... 40
Figure 31: Bar Graph (Gender & Average Score) ... 41
Figure 32: Bar Graph (Geography & Average Score) ... 42
Figure 33: Bar Graph (Qualification Level & Average Score) ... 43
Figure 34: Bar Graph (Employment & Average Score) ... 44
Figure 35: Linear correlation values ... 47
Figure 36: Linear correlation matrix ... 48
Tables
Table 1 ... 45
Acknowledgement
I would like to express my first and foremost gratitude to my thesis advisor Dr. Amir Dirin who not only gave me an important opportunity to write thesis in this topic, but also has been extremely helpful and motivating during the whole thesis process. By virtue of him, I am able to learn something new which is an important milestone for my upcoming future.
I would also like to extend my thanks to Riitta Blomster, my academic advisor who has been always there for me whenever I need any form of support during my studies.
To all the participants that took part in this thesis survey and finally my wife Sirjana Neupane, who supported me throughout my thesis so that I don’t have to worry about other things and could only focus in writing my thesis.
You all have been truly amazing. Thanks a lot.
Kiran Poudel 14 May 2021 Helsinki
1 Introduction
There are more than four hundred thousand peoples with foreign background currently living in Finland. Out of those, many were born abroad and a few of them were born in Finland. Despite being born in Finland, the second generation, whose parents are the ones who were born in foreign countries are also considered to be of foreign backgrounds.
(Statistics Finland 2019)
As per the data as of July 2019, there are 124,396 foreign workforces in Finland. The workforces composed of those who are under working age only. Out of the total workforces, 29,504, which results in 23.7% of the total workforces were unemployed as per the same data. This number is relatively high because the total unemployment rate in whole Finland is only 6.2%. This data shows how difficult it is for foreigners to obtain a job in Finland.
(Foreigner.fi, 2019) However, not only in the case of unemployed foreigners who are financially and emotionally affected by it, but the situation with the employed ones is also not better as compared to their Finnish counterparts. Underpaid, unable to get a permanent contract, overexploitation of labour, not getting qualified jobs are among many reasons leading to dissatisfaction for the foreign employees. (Wall 2019)
There can be various factors that results in life and job satisfaction among the employees.
A research was conducted to determine life satisfaction among various people of Finland, Estonia and The UK based on the marital status of the participants. The aim was to find out the life satisfaction between married and divorced couples. Various results were obtained from the research. Based on gender, women were found to be more satisfied than men.
Also, married ones responded to be more satisfied than divorcees and having a job leads to more satisfaction than not having one. Also, it was found that irrespective of gender and marital status, one having a professional career seemed to be more satisfied than the one who is doing unskilled jobs. (Schoon, Hansson & Salmela-Aro 2005)
As per as job satisfaction is concerned, research conducted by the University of Jyväskylä determines that the dimensions responsible for job satisfaction are mainly skills, work effort, autonomy, job (in)security and payment (Hartikainen, Anttila, Oinas & Nätti 2010).
The thesis mainly targets the foreign nationalities who are currently in Finland and are either working or not working too. This thesis will take into considerations various factors which may lead to life as well as job satisfaction, classify them and analyze them according to the classification. For that, we will use the KNIME analytics platform to study about self and job satisfaction of foreign workers and make different observations based on the analysis.
2 Research Questions and Research Methods
The research question for this thesis to find out the job and self-satisfaction of foreigners working in Finland using the Knime Analytics platform. For this purpose, various patterns will be identified among the participants based on social, failure, entitlement and emotional factors.
Satisfaction is a completely psychological term that is directly and indirectly related to various factors. We will try to look into different factors that overall affects the self and job satisfaction and based on those factors try to find out what are the major factors responsible for the self and job satisfaction for the foreigners who are currently working inside Finland.
Based on that, we can also make suggestions to improve the satisfaction level for the foreigners working in Finland.
In this research the main research questions are:
1. What are the factors that affect the self and job satisfaction level among the foreign workers working in Finland?
2. Is there any related or common pattern among various participants in the emotional schema?
To get the answers, we will look into these sub-questions:
1. What are the major satisfaction factors among the graduate and undergraduate?
2. What are the major dissatisfaction factors in the context of the continent?
3. What are the various common factors among genders, nationalities, qualification level, job status and others?
To get the answers, a survey is conducted among different foreign nationals who are currently living in Finland. The questionnaire is completely anonymous, and the data is used merely to assess and predict foreign employees' job satisfaction in Finland. For this, we have used the Young schema questionnaire. Some more insights into the schema questionnaire can be found on the website, schematherpay.com.
For the research, the Knime analytics platform will be used to construct a project with all the data obtained and later those data are analyzed and discussed.
The research method is quantitative. A quantitative method is that where data are transformed into numerical values and thus analyzed. Even factors like emotions,
measured with numerical values are some of the methods of quantitative research method (Farnsworth 2019).
3 Related Research
Data is a collection of facts that can be a number, words, observations which when processed becomes a piece of information. Data can be structured or unstructured.
Structured data is one that a computer or machine can process. On the other hand, unstructured data can only be processed by human beings. (import.io 2018)
Data is collected and analyzed for various purposes. Those purposes can be related to business, advertisements, government plans and executions and more. Data helps us to provide information, identify problems, give us hidden insights which in turn can be used to make strategies and solutions for the related problems.
The significance of data has led to another term called data mining. The process of discovering co-relation and patterns inside the data is called data mining (Smallcombe, 2020). Data mining is highly applied in the field of business, science, and various research purposes. Data mining not only gives the plain information related to data but also helps to predict future outcomes also. The advancement in computers, artificial intelligence, machine learning and technologies has made data mining more useful and accurate.
Various software is available which can do the data mining task easily.
Self and job satisfaction has been investigated for a long time, especially after globalization where there was more mobilization of people in search of study, jobs and business opportunities.
According to research conducted by Eurofound (European Foundation for the improvement of living and working conditions) inside European unions, it was observed that migrant employers have higher unemployment rates and the ones who are employed are in unskilled jobs where most of them are overqualified (Ambrosini & Barone 2007).
Another study conducted in Taiwan among 440 Thai workers working in Thailand found that there is a positive relationship between job characteristics and job satisfaction which means overqualified jobs may lead to job dissatisfaction for many (Hsu & Liao 2015).
A satisfied employee is not only satisfied within themselves but also add various values to the company such as productivity increment, more loyalty, better teamwork, and high- quality services due to the satisfied employee. On the other hand, less satisfied employee results in less energized work which ultimately results in a decrease in productivity and
There are various methodologies used for data analytics. Waterfall methodology is one of the oldest methodologies used for data analytics (Tutorialspoint). Given the nature of data that evolve in future, more methods were introduced for better data analytics. Agile, Scrum and Kanban are some of the methodologies that are being used these days in place of waterfall methodology (Lucidchart). But along with data mining and the inclusion of artificial intelligence (AI) and machine learning as a part of data analysis, different methodologies were needed.
Some of the popular data analysis methodologies that are popular for data mining are Knowledge discovery in database (KDD), Team data science process (TDSP), and Cross- industry standard process for data mining (CRISP-DM). All these methodologies have their workflow process that involves various phases and have own merits and demerits. (Saltz 2020)
3.1 CRISP-DM
For the research, the CRISP-DM methodology is being used as it is a widely and already proven method for the data mining workflow process. CRISP-DM methodology was introduced in 1966 for the data mining process. As shown in figure 1 below, there are six steps involved during the workflow process of this methodology. They are Business Understanding Data Understanding, Data Preparation, Modeling, Evaluation and Deployment (Rodrigues 2017a).
Figure 1: CRISP-DM Methodology (Vorhies, 2016)
Business Understanding:
The first step in this methodology is business understanding. Without the understanding of data, no project will achieve its desired goals. This is a very important step as it will give context to the data and goals. This step comprises meetings, documentation reading, preparing questions about the context. As the goal of the product should be defined beforehand, thus the development team must have a better understanding of the project.
(Rodrigues 2017b)
Data Understanding:
After the business understanding, the next step is data understanding. The main objective of this step is to find out what can we expect and get from the data. For this purpose, it is necessary to check the quality, competence and trustworthiness of the data. Only if the data possess those qualities, then only the final results can be trusted. In this step, the team members try to extract the best possible information from the data. If any information is not clear, the process steps back to the business understanding to find the relevance of the data in the project. (Rodrigues 2017c)
Data Preparation:
The third step is data preparation and in most of the project, it is the most time-consuming step. Because of it, it is also one of the most important steps. As there could be a lot of data in some cases, this process could get complex in some cases also. For example., there could be a situation where string data has to be converted into numeric data and vice versa depending on the situation. For that, the development team will work into that so the data can be normalized into the desired format. (Rodrigues 2017d)
Modeling:
This part is the core part of the project which provides results for the project. However, being a core process, it is a less time-consuming process if previous steps have been done correctly. If the required results are not achieved, the step goes back to data preparation again. (Rodrigues 2017e)
Evaluation:
yield the desired outcome. However, there could be some error and the range of error could define the validity of the result. A smaller error is quite common and acceptable, but if the error is huge, then this step can go to the first step again. (Rodrigues 2017f)
Deployment:
This is the final step of the process, and this is also the only step that is not a part of the process as the step doesn’t follow the reverse path. In this step, the outcome of the results is presented in a useful and understandable way. (Rodrigues 2017g)
3.2 Data Mining With Knime
Konstanz Information Miner or simply KNIME is an open-source software specially designed for data analysis and creating data science. One of the main features of this software is that not only it does data analysis from the input data, but also it has machine learning features that provide various extra functions like regressions and predictions. Adding to that, it can also create a visual representation of various data so that it is easy to understand for other users also. (Breaker 2019)
Though there is various software for data analysis, Knime is one of the most used data analysis software. One of the main reasons for its popularity is that Knime is an open-source software and is completely free to use, which makes it easily accessible to many users who are looking for a better data analysis tool without spending a lot of money. Using Knime is also relatively easy as it follows the drag and drop method for using it. Apart from it, its machine learning algorithms make it an even better tool for data analysis. Various algorithms such as decision tree, random forest, linear regression, polynomial regression, logistics regressions and predictions also included in Knime. (Silipo 2020)
Thus, we found out Knime analytics platform to be the ideal software for data analysing in the case of our research.
3.3 Young Schema Questionnaire
Young Schema Questionnaire (YSQ) was first developed in 1990 (Pauwels, Claes, Smits et al 2013). Some changes were introduced from time to time and the current version of YSQ is the 3rd edition. The YSQ has both long-form and short form. There are a total of 18 early maladaptive schemas in YSQ. (Specific inventories)
However, in this thesis, we will be only working out four of them which are: social isolation, emotional deprivation, entitlement and failure to achieve. The questions require answering
on what the participant feels emotionally rather than what they think to be true. The answers options are in Likert-scale form ranging from completely untrue of me to describes me perfectly. (Schema Therapy)
3.4 Statistical vs Data Mining
Statistics and data mining are often confused to be similar, but they are different but interrelated. Statistics mainly refers to the analysis, interpretation and presentation of data whereas data mining is the process of extracting various information from the data. (Mayo)
Statistics is the core of data mining. In statistics, data are in numeric form, whereas in data mining, data can be either numeric or non-numeric also. Also, data mining is suitable where the data sets are large as statistics are only suitable for small data sets. Statistics is a deductive process that doesn’t make any kind of predictions and data mining is an inductive process where new predictions can be made. One of the main advantages of data mining is also that it uses heuristics think which is a rule to form judgments and make decisions which in the case of statistics don’t have. (Pedamkar) Thus, for this thesis, we have used a data mining tool for its advantage over statistics.
4 Empirical Study/Designing the Research
The questionnaire was prepared by the thesis supervisor in consultation with psychologists.
A young schema questionnaire was used to prepare the questions. (Young schema questionnaires: Long and short forms) The questions are composed of basic information such as country of birth, gender, qualification and employment status. No personal information such as name, address or contacts were asked to make the data anonymous and to maintain the privacy of the respondents.
Apart from those basic questions, 35 questions were asked on different topics. Those questions were mainly designed to understand various aspects of the work and life situation of the respondents and how they feel about those. Instead of a numeric response, strings were used for the responses where the given options were: Completely untrue of me, mostly untrue of me, slightly more true than untrue, moderately true of me, mostly true of me and describes me perfectly. The use of the same options for each question also means that if needed for the responses to be converted into numeric form, it would be easy.
After the questions were set, a google form was created and links were shared in various social media groups to collect the data. Emails, Facebook, WhatsApp are some of the examples where the link was shared. Various participants filled the form and when we have collected our desired number of data sets, data were presented in an excel file that contains all the responses provided by the participants and then the Knime analytics tool was used for further analysis.
5 Data Analysis
5.1 Data Preparations
When the intended number of data was collected, the thesis supervisor transformed the data into an excel sheet and then it was forward to the thesis writer. The purpose of containing the collected data into an excel file was to make the data readable for the user as well as for Knime.
Knime supports various types of files such as excel table and CSV. We choose to contain the data into an excel file so that we can use it in our analysis tool.
For the data analysis using Knime, various nodes were used. Some of the important nodes that were used are explained below:
Excel Reader:
As the data was in excel format, the excel reader node was used as the very first node to read the data. The purpose of file reader is the very first step used in Knime analytics, only after that, other functions can be carried out.
Figure 2: Excel Reader Node and its Configuration
Here in above figure 2, we can see the excel reader node in Knime and the configuration window. We can browse and then load the excel file in the configuration window and if the file is correct, we can also see the preview at the bottom of the configuration window as shown in the above picture. After loading the data in the form of an excel file, then we can analyse the data in as different ways as we want.
Row and Column Filter:
Row and column filter nodes are mainly used to filter data from rows or columns. This is very useful especially when we want to analyse only certain elements of rows or columns.
Knime analytics tools support row and column filter with row filter and column filter nodes.
We can configure these nodes and filter the data by inclusion and exclusion based on various parameters such as attribute value, number and row id. The row filter node and its dialogue window are as shown in figure 3 below.
Figure 3: Row Filter
Similarly, a column filter can be used to filter various data in the column as shown in figure 4. We can include and exclude different columns by easily putting them into the inclusion or exclusion box as shown in the picture below.
Figure 4: Column Filter
Rule Engine:
The rule engine can be used for considering various user-defined rules for the data. We can put more than one rule for the same file at a time. Each rule is represented by a line and after that condition and the outcome based on the condition is defined. When the defined rule matches the condition, a new column is created with the outcome value.
For our data analysis, the rule engine node was used in a different situation to classify data into new forms. E.g.: as we can see in the below-shown figure 5, we have used a rule engine to create a new column that classifies the participants into the employed or not category based on their employment status scenario.
Figure 5: Rule Engine Condition
Here we have mentioned a condition about employment status to include internship, student and unemployed as No in terms of employed or unemployed classification. For the rest of the conditions, it will hold Yes, as written above in the expression window of the rule engine.
After this condition is applied, a new column with column name Employed or Unemployed will be added to the data as shown in figure 6 below:
Figure 6: Addition of New Column in Rule Engine
Pivoting:
Pivoting nodes are especially used to view data from a different perspective. In our data analysis, we have used pivoting nodes to compare two different sets of information and the relation between them in terms of various factors.
One of such examples which we have used in our data analysis is to compare the qualification based on gender. As the initial data also contain that information, but it was not visible at a glance. Thus, we used pivoting node to compare gender and qualification and find out the relation between these two factors.
Figure 7: Pivoting
As we can see in the above figure 7, we used pivoting node to create a pivot table for gender and qualification rank which can also be represented in a visual form as shown in the picture above. In a similar way pivoting nodes were used to compare various data during the data analysis which will be discussed more in the result part of the thesis.
String to Number:
The string to number node is used to convert the strings in a column into numbers. As we can see from figure 8 below, we can choose the columns which data set we intend to convert into numbers. And based on those selections, the type of data in those columns are changed from string to numbers.
Figure 8: String to Number
Math Formula:
Math formula node is a very important node which can perform various mathematical expression based on the values in a data set. Here we can see from the below picture 9 that, math formula can be used for various functions such as row count, to find minimum, maximum, mean value and so on.
Figure 9: Math Formula
Linear correlation:
This node measures the correlation between two different variables. As shown in below figure 10, we can choose different variables for which we want to measure the correlation.
Based on that selection, a correlation value is calculated which ranges from -1 to +1 where -1 means strong negative correlation and +1 means strong positive correlation. Numbers falling between them can be categorized as weak and moderate correlation.
Figure 10: Linear Correlation
6 Results
Upon data analysis, various results were obtained. In this chapter, we will look over those findings. We will classify results into three different factors; Participant profile, comparisons and predictions and explain them here.
6.1 Participants Profile
As many as 90 participants took part in the survey. The participants were mostly people with a foreign background other than Finland and are living in Finland. There were also 2 participants from Finland and 2 other participants whose country of birth was not mentioned.
However, we have included these 4 participants also for the analysis of data.
Discussing the country, there were participants from as many as 42 different countries which is almost over 2 persons from each country on average. However, the actual number of participants varies with different countries.
As shown in figure 11 below, the highest number of participants were from Nepal which was 12 participants in total. It was followed by Iran with 10 participants, India, Vietnam and USA with 5 participants each, Ukraine with 4 participants, UK, China and Estonia with 3 participants each, Russia, Portugal, Kenya, Finland and Ethiopia with 2 participants each.
All other remaining countries: Argentina, Bangladesh, Belarus, Brazil, Cameroon, Czech Republic, Egypt, France, Germany, Ghana, Hungary, Indonesia, Iraq, Ireland, Italy, Jordan, Kosovo, Lithuania, Netherlands, Peru, Philippines, Poland, Russian Federation, Singapore, Somalia, Switzerland, Thailand and Turkey has one participant each. And as mentioned already before, two participants country of birth was not mentioned, thus we mentioned NA which stands for Not Available in the country of birth section for those participants.
Figure 11: Pie Chart (Country)
Because there were participants from several different countries, we classified participants according to geographical distribution, which in our case we used continent as geography for ease of data analysis. In this classification, we have also included Eurasia as different geography for those countries which falls in both Asia and Europe.
As we can see from the above-mentioned figure 12, upon classification based on geographical distribution, the majority of participants were from Asia where there were as many as 42 participants. The second highest participants were from Europe with 26 participants, and it was followed by Africa with 8 participants, North America with 5 participants, Eurasia with 4 participants and South America with 3 participants.
If we measure these numbers in terms of percentage, we will get 47 % of participants from Asia, 29% from Europe, 9% from Africa, 6% from North America, 4% from Eurasia and 3%
from South America. And as mentioned earlier, two participants country of birth is not available, thus the geographical classification was also not available for those two participants. The overall distribution percentage after the classification of countries according to their geographical location is as shown in figure 13 below.
Figure 13: Pie Chart (Continent in Percentage)
Apart from the country of birth and respective continents, the participants were classified on several other factors such as gender, qualification levels and employment status.
As we can see in figure 14, out of 90 participants, 46 were male and 43 were female. One of the participants identifies as non-binary. Similarly, in terms of percentage, the percentage of the male and female participant was 51 and 48 percent respectively.
Figure 14: Pie Chart (Gender)
When asked about their educational qualification, most of the participants mentioned having completed either bachelor’s degree or a master’s degree. Few other participants were students or doing their PhD. The numbers are also shown in figure 15 below.
We further make our classification regarding qualification level, where we classified the education qualification into a diploma, undergraduate, graduate and post-graduate.
Participants with education qualification below bachelor’s degree were termed as diploma, those completing bachelor’s degree were termed as undergraduate, those with master’s degree were termed as graduate and those doing PhD were termed as post-graduate.
Based on that classification, it was found that 9 participants had the diploma-level qualification, 46 were undergraduate, 33 were graduate and only 2 of them were post- graduate. The overall numbers are shown in the above-mentioned figure 16.
Figure 16: Pie Chart (Qualification Rank)
And finally, the final classification of the user profile was made based on the employment status. Participants were asked about their employment status, and it was found that as many as 52 participants were employed full time and 14 were unemployed. The remaining participants mentioned being a part-time employed, entrepreneur, business owner, self- employed, PhD student, doing internship and students. The representation is also shown in figure 17, as shown below.
Figure 17: Bar Graph (Employment Status)
As in qualification classification, we also made similar qualification of employment status to find out if the participants are employed or not. In this classification, any sort of employment, full time or part-time, entrepreneur, business owner, self-employed were considered to be employed and, students and those doing internships were considered to be unemployed.
However, we have classified PhD students to be employed in our classification.
Doing so, we found that 72 participants have some sort of employment that generates income and the remaining 18 were found out to be not employed based on their responses.
The numbers are shown in the below figure 18 also.
Figure 18: Pie Chart (Employed or Not)
In terms of percentage, those who are employed accounts for 80% of the total participants and the remaining 20% are not employed.
6.2 Comparisons
Various comparisons were made to understand the relation between more than one factors.
Those comparisons were made to view data in a different perspective and thus pivoting nodes were used to make these comparisons and bar graphs were used to display the comparison results in graphics. For comparisons, we have used factors such as gender, qualification level, employment status and geographical distribution i.e., continent.
First comparisons, we made regarding gender with qualification and employment status. As shown in the below-shown figure 19, we found out that among male and female participants, were equally distributed as a similar number of males as well of females has similar qualification level. As there were 46 males and 43 females in the survey and out of those numbers, 5 had the diploma-level qualification, 22 undergraduates, 18 graduates and 1 post-graduate against 3 females who had a diploma, 24 undergraduates, 15 graduates and 1 post-graduate qualification. From this, we can say that gender plays a minimum to no role against the qualification level as we found that both male and female participants were equally qualified.
Figure 19: Bar Graph (Gender and Qualification Level)
Then gender and employment status were also compared with each other. It was found that despite having a bit few numbers of female in the survey as compared to the male which was 43 females and 46 males, a bit more females were employed as well as bit few females were unemployed as compared to male. In terms of employment, 36 females were employed as compared to 35 males who were employed. However, as many as 11 males were unemployed as compared to females whose number of employments was only 7. This is shown in figure 20 in form of the bar graph. Thus, according to this survey more females are employed as compared to males, though it is worth saying that the difference is very minimal.
Figure 20: Bar Graph (Gender and Employed/Not)
The next comparison was made on basis of geographical distribution concerning qualification and employment status. As shown in figure 21, among the 46 undergraduates, 22 are from Asia, 10 from Europe, 6 from Africa, 3 from Eurasia, 3 from North America, 1 from South America and 1 from those participants whose country of birth was not mentioned. Similarly, among 33 graduates, 16 were from Asia, 11 from Europe, 2 from Africa, 2 from North America and 2 from South America. And for 9 diploma-level qualifications, 5 were from Europe, 2 from Asia, 1 from Eurasia and 1 was not mentioned.
However, all of the post-graduates which number stands at 2 were from Asia only which mean none of the other participants except those from Asia had or is pursuing a post- graduate degree.
Figure 21: Bar Graph (Continent and Qualification Level)
Here we have to consider that the number of participants in the survey was varied according to each continent where Asia and Europe have the highest number of participants in the survey, thus the number for these continents tends to be high than other continents.
AND
Figure 22: Pie Chart (Percentage of Graduate and Under-graduate in different Continents)
As our survey consists of 47% Asian, 29% Europeans, 9% Africans, 4% Eurasians, 6%
North Americans, 3% South Americans and rest as not mentioned ones, the qualification
level according to the different continent was compared. As shown in above figure 22, we made percentage comparisons of graduates and undergraduates in different continents and found that the total percentage of Asians among the graduates and undergraduates was 48% each, which is similar to the 47% of Asian who participated in this survey. Europeans with 29% population in the survey had also similar percentage among the graduates and undergraduates which were 33% and 22% respectively. The ratio was similar with people from Africa, North America as well as North America. Diploma and postgraduates were not compared in this comparison as they make a very small number in terms of qualification levels.
As now know that the ratio of the participant from the different geographical area following their qualification is similar, we can now observe other factors and then compare against each other such as comparison of employment status according to geographical distribution. Thus, we compare employment status with different continents and the observation is displayed in below figure 23.
Figure 23: Bar Graph (Continent and Employed/Not)
Next, we try to compare these numbers with actual geographical distribution. During those comparisons, we observed that the ratio of employment and the population distribution is
and 26% share in terms of population and employment. The difference here is not big enough to observe any vital contrast among participant’s geographical distribution and chances of them being employed. The overall percentage figure of employed ones from different continents is shown in figure 24 below.
Figure 24: Pie Chart (Percentage of Employed in Different Continent)
We made a similar comparison with those who are not employed and this time we observed a very different result. Here, as shown in figure 25, we found that Europe despite having fewer participants than Asia has more unemployed ones than Asia which was 39% against 33% of Asia. Thus, Asia despite having almost 50% participants only contributed to one- third of unemployment and Europe has 10 percentage point more people in terms of unemployment as compared to their population in the survey. The number of unemployed ones from North America and Eurasia was zero. And South America contributed to 11%
unemployment despite having only 3% participation in the total survey.
:
Figure 25: Pie Chart (Percentage of Unemployed in Different Continent)
And finally, we also made a comparison between qualification level and employment status to analyse the role in which qualification plays a role in employment or not. We found out that out of 60 undergraduates 48 were employed against 12 who were not employed. This in percentage turns out to be 80% employed and 20% not employed among undergraduates.
Similarly, for graduates out of 28, 22 were employed and 6 were unemployed which again converting into percentage gives us 78.6% employed and 21.4% unemployed among the graduates. Thus, the chance of getting employed for graduate and undergraduate are more or less similar with undergraduate ones slightly holding a narrow age in terms of percentage.
As we have already included post-graduate ones to be employed ones, so they make 100%
employed. The overall numbers are displayed in form of a bar graph as shown in below figure 26.
Figure 26: Bar Graph (Qualification and Employed/Not)
6.3 Findings
In this part of the result, we will analyse the objective answers asked the participants and analyse them using the Knime analytics tool. There was a total of 35 objective questions asked to the participants and each question was Likert scales. The scales were completely true of me, mostly true of me, slightly more true than untrue, moderately true of me, mostly true of me and describes me perfectly.
Based on the discussion with thesis advisors, we classified questions into four factors was Social isolation, Failure, Entitlement and Emotional. Not all questions were considered into this classification and also some questions were put in more than one factor also if they met the criteria for those factors. A total of 4 questions were grouped for each factor.
Social Isolation:
• I feel alienated from other people
• I don’t belong, I’m a loner
• I’m fundamentally different from other people
• I don’t fit in
Failure:
• I’m not as talented as most people are at their work
• Most other people are more capable than I am in the area of work (or school) and achievements
• I’m incompetent when it comes to achievement
• Almost nothing I do at work (or school) is as good as other people can do
Entitlement:
• I felt that I shouldn’t have to follow the normal rules and conventions other people do
• I have to be constrained or kept from doing what I want
• I’m fundamentally different from other people
• I’m special and shouldn’t have to accept many of the restrictions placed on other people
Emotional:
• For the most part, I’ve not had someone who listens to me, understands me, or is turned into my true needs and feelings
• For much of my life, I haven’t felt that I am special to someone
• In general, people have not been there to give me warmth, holdings and affection
• Most of the time, I haven’t had someone to nurture me, share him/herself with me, or care deeply about everything that happens to me
As the answers were in string form, we converted the answers into integers using rule engine and string to integer node in Knime. The reason we have to apply string to integer node even after converting the answers into a numeric form using rule engine is that even after applying rule engine, the data is still considered to be in string form, so we have to use an extra node to convert those numbers again into integer form.
The numeric scores given for each answer were:
Completely untrue of me 0
Mostly untrue of me 1
Slightly more true than untrue 2 Moderately true of me 3
Mostly true of me 4
Figure 27: Answers in string form
Figure 28: Answers in numeric form
We can see, in the above figure 27, the answers were in string form and after conversion, the answers were transformed into numeric form as shown in figure 28.
After the answers were transformed into a numeric value, the next step was to group the questions for different factors. As mentioned earlier, 4 different factors: social isolation, failure, entitlement and emotional were defined and each factor consists of 4 questions.
Now, as we have answers in numeric form, we add these numbers for each factor and then calculate an average value. For e.g., if each of the four questions in the factor social isolation has an answer of 1,2,3 and 4 respectively for a user, the average score of social isolation for that user will be (1+2+3+4)/4 which will be 2.5 which is somewhat between slightly more
true and moderately true. Similar process was conducted for remaining factors also and doing so, we get an average score of each user. We used a math formula node to calculate these numbers.
Figure 29: Average score of each user for different factors
As shown in figure 29 above, we can see that the average score of social isolation for the user in row 0 is 3.5. Similarly, he/she has a failure average of 1.5, entitlement average of 2 and an emotional average of 2.25.
Next, we calculate the mean value of these 4 different averages using the math formula node. This gives us the average score among all user for a different factor. As shown in below figure 30, we have the social isolation mean of all users as 1.6, failure mean as 0.956, entitlement mean as 1.919 and emotional mean as 1.375.
As the scoring ranges from 0 to 5, we have to understand that for each factor, a number represents a different story. In the case of social isolation, a low score means the person felt less socially isolated, whereas a higher score means that the person is more socially isolated. Next for failure factor, a low score means the person is less afraid of failure or doesn’t consider him/herself inferior, less talented or an underachiever. And a higher score in failure means the person is more afraid of failing, consider him/herself not talented and doesn’t believe in his ability. Similarly, for the entitlement factor, a low score means the
some sort of emotional support for him/her in time of needs and a high score means the opposite of it.
Figure 30: Mean score of all users
Now compared to all users we have got the mean score for all four factors. This mean score can be used to compare different users and groups if they have more or less score than the mean score.
Based on the average score of each user, we classify users based on gender, geography, educational level and employment status. The results are displayed in form of a bar graph as shown in figure 31, 32, 33 and 34 respectively.
Figure 31: Bar Graph (Gender & Average Score)
Figure 31 displays the average score for social isolation, failure, entitlement and emotion with respect to genders. As we can see from the graph, the social isolation score is highest for non-binary, male and female respectively. Next, the failure average is equal for male and non-binary and a tad less in the case of female. Similarly, the entitlement average is highest among males among all three genders and so is the emotional average also.
Figure 32: Bar Graph (Geography & Average Score)
Figure 32 displays the average score for social isolation, failure, entitlement and emotion with respect to geography. Barring the not mentioned ones, we can see from the graph that the social isolation score is highest for North Americans and lowest among Asians. The failure average is again highest for North Americans and lowest among Africans and South Americans. Next, the entitlement average is highest among North Americans and lowest among South Americans. Finally, for the emotional average, Europeans have the highest emotional average as compared to others where Africans have the lowest.
Figure 33: Bar Graph (Qualification Level & Average Score)
Figure 33 represents the average score on basis of four different average scores as compared to qualification level. As we can see, the social isolation average was highest among the post-graduates and lowest among graduates. Next, the failure average is highest for Diploma holders and lowest for post-graduates. Then, the entitlement average was again highest for the post-graduates and lowest for diploma and graduates. And finally, the emotional average was highest for post-graduates and lowest for graduates.
Figure 34: Bar Graph (Employment & Average Score)
Figure 34 displays the average score based on different factors as compared to employment. We can see that people who are not employed have higher social isolation score as compared to those who have a job. Similar is the trend in case of failure average and emotional average also. However, the entitlement average was the same among both.
Then we also created a table containing all of these average scores so that we can make some observations from the table. The table also contains the average scores of each of the four factors which we have already calculated as shown in figure 30 above also.
Factors (Average Score)
Groups
Social Isolation (1.60)
Failure (0.96)
Entitlement (1.92)
Emotional (1.37)
GENDER
Male 1.68 1.00 2.08 1.48
Female 1.48 0.91 1.75 1.27
Non-binary 3.00 1.00 2.00 1.00
GEOGRAPHY
Asia 1.34 0.95 1.85 1.35
Africa 1.41 0.75 2.06 1.12
South America 1.75 0.75 1.33 1.17
North America 2.20 1.10 2.10 1.20
Europe 1.95 1.00 2.06 1.51
Eurasia 1.50 0.94 1.44 1.37
Not Mentioned 1.75 1.37 2.37 1.87
QUALIFICATION LEVEL
Diploma 1.50 1.61 1.72 1.36
Under-graduate 1.64 1.02 2.01 1.46
Graduate 1.44 0.71 1.71 1.16
Post-graduate 3.75 0.62 4.12 3.00
EMPLOYMENT STATUS
Employed (Yes) 1.55 0.86 1.92 1.30
Not Employed (No) 1.79 1.33 1.92 1.65
Table 1
As we can see from the table above, based on gender, male felt that they are more socially isolated as compared to female. However, in case of failure, females have less failure score as compared to male as they believe they don’t feel inferior as compared to others. For entitlement, male felt more entitled as compared to female and finally for emotional support factor female felt that they had more emotional support in their life as compared to male.
We also have only 1 participant whose gender is non-binary. Though 1 might be a bit few number to make analysis, but based on their answer it was observed that non-binary felt
they have more emotional support in their life as compared to the other two genders. So, we can say that gender plays a role in these factors where female felt less social isolation, has less failure factor and more emotional support as compared to males and males felt more entitled as compared to female.
In the case of geographical classification, it was observed that Asians felt less socially isolated followed by Africans, Eurasians, South Americans, Europeans and North Americans respectively. This might be also the result of the social structure of these geographies where Asians and Africans have a close-knit society as compared to Europeans and North Americans where individuality and independence are more important.
In case of failure, Africans and South Americans have less failure factor as compared to others. In terms of entitlement, North Americans, closely followed by Europeans and Africans felt more entitled as compared to others. And for the emotional factor, Europeans had a higher emotional factor score as compared to others which means that they felt that they don’t have much emotional support in their life and times of needs as compared to others. Africans and South Americans, on the other hand, has the lowest emotional support factors meaning they have more support from other people in their life. Thus, on basis of the observation, we can say that in geography wise we have different scores. The social isolation, failure, entitlement as well as emotion varies based on different geographical location.
When observed on basis of qualification level, we found that among, diploma, under- graduates and graduates, graduates felt less socially isolated the among other two. Next, on the failure factor, again graduates have less failure factor compared to the other two.
However, the entitlement factor was high among undergraduates and similar between diploma and graduates. And finally, the emotional factor was least for graduates which means they claim to have more emotional support in their life as compared to others. In this classification based on educational level, we compare post-graduates solely because they only had two participants. However, interesting observations were observed for post- graduates where if compared among all, they felt they are more socially isolated, had the least failure factor, the had highest entitlement factor of as high as 4.12 and they also had the highest emotional factor score which means they had very less emotional support in their life. This even after having a low failure score and higher entitlement score, post- graduates claimed to be more socially isolated having no emotional support from others.
And finally, on basis of employment status, we observe that employed one has less social isolation score as compared to unemployed one and they also have less failure score as compared to unemployed one. Though the entitlement score was same for both, the emotional score was less for employed ones which mean they have more emotional support
than the unemployed ones. This finding was really interesting as how having a job plays a vital role in social life, low fear of failure and also more emotional support.
6.4 Correlation
In this section, we will be discussing the correlation between social isolation, failure, entitlement and emotional factor and try to find out how strongly or weakly they are correlated. This also gives us an estimation about how a change in one of the factors will have an impact on others. For this purpose, we will be applying linear correlation between these results and based on the results we can define the correlation between them. In our findings, we will consider values between above 0 and below 0.3 to be weak positive relation, between 0.3 to below 0.7 to be moderate positive relation and between 0.7 to below 1 to be strong positive relation.
Before that, we have to understand that the correlation coefficient ranges from -1 to +1 where a value of +1 means a perfect positive relationship. 0 means no relationship and -1 means perfect negative relationship.
Figure 35: Linear correlation values
The correlation of different factors with each other is as shown in the above figure 35. We can see that the correlation value for social isolation with failure, entitlement and emotional averages is 0.41, 0.54 and 0.58 respectively, which means that social isolation has a moderate positive relationship with failure, entitlement and emotional.
Similarly, failure has a weak positive relationship with entitlement and moderate positive relation with the Emotional factor. And finally, both entitlement and emotional factor have a moderate positive relationship with each other.
The finding can also be shown in a graphical form as shown in below figure 36.
Figure 36: Linear correlation matrix
7 Discussions
Data mining is an important tool for decision making. With the help of data mining, one can find various information within the data such as hidden patterns and based on that various decisions can be made. Also, as technology progresses, data mining has been easier as of today than in past. Various data mining tools are available which not only helps to make decisions in present but also can give future predictions so that one can be prepared for the future also. In the case of an organisation and its employees also, data mining is very important. The organisation can learn about their employees such as how they feel as an individual and also in their jobs and based on that, they can offer help to the employees. A satisfied employee both in their life and their job ultimately means a beneficial point for the organisation as it increases the efficiency of the workforce within the organisation.
Now, let’s discuss the research questions and their answers.
7.1 Research Question’s Answers
Our main research questions were:
1. What are the factors that affect the self and job satisfaction level among the foreign workers working in Finland?
2. Is there any related or common pattern among various participants in the emotional schema?
To get these answers, sub-questions were created, and they are discussed as below:
1. What are the major satisfaction factors among the graduate and undergraduate?
When we compare various factors on basis of qualifications, we found that graduates are more satisfied in the context of social isolation where they felt less socially isolated as compared to undergraduates. The satisfaction level was highest among the post-graduates in terms of social isolation. Similar was the observation when it comes to fear of failure as well as emotional support, where graduates enjoy less fear of failure and more emotional support in their lives as compared to undergraduates. (Figure 33)
However, it was worth noticing that under-graduates felt more satisfied in terms of entitlement as compared to graduates (Figure 33). This means undergraduates
Thus, we can say that major satisfaction factors among graduates were less social isolation, low fear of failure and more emotional support compared to undergraduates and on the other hands, major satisfaction for undergraduate was entitlement where they average higher than the graduates.
2. What are the major dissatisfaction factors in the context of the continent?
When we make comparisons and observations as compared to geographical location, we found that Europeans and North Americans are more dissatisfied with their social life and feeling of being alone and not fitting in as compared to people from other continents. We also observe that the fear of failure is also high among the people from these continents as compared to others. (Figure 32)
On the other hand, South Americans felt they are less entitled as compared to others. And, for emotional support, Europeans are more dissatisfied as they experience less emotional support in their life as compared to others. (Figure 32)
3. What are the various common factors among genders, nationalities, qualification level, job status and others?
Looking into the results, we observed that for both male and females, the average score in terms of social isolation, failure, entitlement and emotional were higher for males as compared to female. However, it is worth mentioning that each score higher and lower have different meaning where in some, higher is better and in some lower score is considered to be better. If we rank those scores also, they were also similar in the case of both genders as the rank of all scores for each gender was the same. As we can see from table 1, the highest average score for male was 2.08, which was of Entitlement, followed by social isolation, emotion and failure with scores of 1.68, 1.48 and 1 respectively. A similar thing can be observed in the case of the female also where the score ranks in the same order as of male. (Table 1)
In geographical distribution according to different nationalities, the failure average score was same for both Africans and South Americans, whereas the entitlement score was same for Africans and Europeans. (Figure 32)
The entitlement average was almost the same for graduates and diploma holders despite other averages being different. ((Figure 33)
And finally, in terms of employment status, the entitlement average was same for both unemployed and unemployed ones. This is an interesting common factor because if we look for other factors, employed ones are better situated in terms of social isolation, failure and emotional factors. (Figure 34)
7.2 Arguments
Research shows that males have more risk of isolation as compared to females. The observation is supported by the fact that male makes friends less easily as compared to a woman and also participate less in social and community activities. (VAS, 2018) A similar result was observed in the case of our analysis too as discussed in the results (Figure 31).
The next thing that we observed was the failure average was less for female as compared to male, however, the difference was very minimum (Figure 31). But various studies show that the fear of failure is higher for females than males. One of the reasons for it is that females are more likely to be a perfectionist because they tend to take any task when they are 100% sure about their skills and qualifications whereas the percentage for male is only 50% which in a way shows males are more confident than females. (Bennett 2014)
In terms of emotional support, females have expressed that they have more emotional support as compared to male (Figure 31). This observation is also supported by research from a mental health charity named mind that females receive more emotional support as compared to males because males do not open up about their emotional issues as easily as females. (Mind 2016)
As our results show male feels more entitled than female and various studies in the similar topic also support this (Figure 31). An article published in The Guardian how males feel entitled as compared to females despite doing similar job. It further adds that for household chores also males feel that they don’t have to or do less than females. (Bindel 2017)
One of the observations that were made is that Asians felt least socially isolate in comparison to Europeans and North Americans (Figure 32). Several pieces of research support this finding. According to a study, loneliness in Asia, Africa and South America is very rare as compared to Europe and North America. Also, English speaking countries like the USA, UK, Australia and Canada are the loneliest by far. (Denman)
For people with different educational backgrounds, we observed that higher education
find out that the level of education impacts the self-esteem of a person also where the person of higher education has higher self-esteem meaning that the person doesn’t feel that he/she is less talented than other and can’t achieve anything in life. Thus, the higher the education level, the less would be the fear of failure factor as self-esteem increases along with the increase in educational level. (Saygılı, Kesecioğlu & Kırıktaş)
Discussing in terms of employment, a piece of news published in the BBC mentioned that unemployed young people have more chances of being socially isolated. Various people who were part of the news story mentioned being unemployed has led them to lose their confidence, generate low self-esteem and even stopping to talk to friends, all of which can have a direct effect in terms of fear of failure and emotional support also. (BBC, 2015) We have also similar findings as discussed in the results (Figure 34).
7.3 Reliability and Validity
The thesis intends to analyse job and self-satisfaction data using the KNIME data mining tool. Knime is one of the most used tools for data mining and is also relatively easy to use.
On top of that, it is open source and is free to use. This makes KNIME a reliable tool to analyse the data. The data was obtained from various persons living in Finland. A google document was created and was circulated into various groups and pages and based on different people responses, data were collected. Because a google form requires login to fill data and single login can only fill data once, so we can say that all the participants were unique. To make the data anonymous no questions such as name, contact number, emails or addresses were asked in the questionnaires.
Young Schema Questionnaire (YSQ) was used for asking questions to the respondents.
One of the benefits of YSQ was also that it had Likert-scale answers which means the answers don’t have truly positive or truly negative options but has 6 different options to choose from between truly positive and truly negative ones. This helps to understand better emotion of the participants also. Despite collecting the data randomly and anonymously, we were able to get data from people with different backgrounds in terms of gender, geography, education qualification as well as employment status.
The thesis advisor Dr. Amir Dirin had created the questionnaires himself. As the thesis topic was proposed by the thesis advisor himself, various research was done prior to the topic before the start of the thesis. Also, various other researchers are involved in this topic and research to study other dimensions and to find other findings.
The results that were obtained from this thesis was compared to different previous studies and findings and many of our observations were supported by previous studies and findings.
More about this has already been discussed in the argument section of the thesis results.
Thus, we can say that the thesis holds reliability and validity in terms of study and results.
8 Conclusions
The main objective of this thesis was to analyse the job and self-satisfaction data using the KNIME analytics platform. Reliable data were collected and analysed using the platform.
One of the major parts of the thesis was to learn to use the KNIME analytics platform.
Various challenges occurred during the learning as well as implementation phases but were eventually overcome.
Although we were able to collect the intended numbers of data, some of the data collected were short in some categories such as we would have liked to collect more data of post- graduate participants in comparison to what we have collected for the thesis. However, the data was collected randomly so these things are bound to occur. But in future, I would have loved to collect more data for similar studies and then create a random but equal proportion of participants based on various factors and then analyse.
To conclude, we were able to properly use the KNIME analytics platform to analyse the self and job satisfaction data and we hope, this study will be able to prove useful presently and also for future research.
9 References
All Answers Ltd. 2018. Causes and Effect of Job Satisfaction on a Company URL:https://ukdiss.com/examples/job-satisfaction.php?vref=1
Ambrosini, M. & Barone, C. 2007. Employment and working conditions of migrant workers
URL:https://www.eurofound.europa.eu/publications/report/2007/employment-and- working-conditions-of-migrant-workers
BBC 2015. Jobless young people at risk of isolation, charity says URL:https://www.bbc.com/news/uk-30803492
Bennett, J. 2014. It’s not you, it’s science: How perfectionism holds women back URL:https://time.com/70558/its-not-you-its-science-how-perfectionism-holds- women-back/
Bindel, J. 2017. Salma Hayek is right: compared with women, men are lazy and entitled
URL:https://www.theguardian.com/commentisfree/2017/aug/11/salma-hayek- feminism-inequality-men-women
Breaker, D. 2019. Knime analytics: overview and review
URL:https://www.sqlbot.co/blog/knime-analytics-overview-and-review
Denman, G. All the lonely people-the epidemic of loneliness and its consequences URL:https://socialscienceworks.org/all-the-lonely-people-the-epidemic-of-
loneliness-and-its-consequences/
Farnsworth, B. 2019. Qualitative vs quantitative research-what is what?
URL:https://imotions.com/blog/qualitative-vs-quantitative-research/
Foreigner.fi 2019. 23.7% of the foreign workforce is unemployed
URL:https://www.foreigner.fi/articulo/work-and-study/70-of-foreign-workforce- concentrated-in-3-finnish-
regions/20190807174532002586.html#:~:text=70.4%25%20of%20the%20men%20 and,live%20in%20three%20Finnish%20regions.&text=In%20Finland%2C%20the%
20workforce%20of,according%20to%20Statistics%20Finland%20data.
