The participation of women in science, technology, engineering, and mathematics (STEM) programmes, especially in computer science, remains a challenge and a relevant area of study. Understanding the characteristics, motivations, and backgrounds of female students admitted to these disciplines is important for designing support strategies and encouraging them to remain in these programmes. Therefore, this exploratory study aims to identify association rules among various characteristics of female students admitted (N=30) into informatics programs at the Facultad Politécnica of the Universidad Nacional de Asunción (FP-UNA) in the 2024 cohort. Data were collected using a 33-question survey instrument across six dimensions. Subsequently, the FP-Growth algorithm was applied for association rule mining. The findings reveal consistent patterns and direct associations. These preliminary results offer points for discussion on the diversity of incoming students' profiles and demonstrate the usefulness of association rules as a tool for better understanding this student group in the context of the FP-UNA.
higher education, women, gender, computing, educational datamining, fpgrowth 1
At the Facultad Politécnica of the Universidad Nacional de Asunción (FP-UNA), available
data [
Given the relevance of gender equality in technology education, this work aims to better understand the profiles of newly admitted students through a detailed analysis. Specifically, the objective is to identify patterns of association between various characteristics of female students admitted to informatics studies at FP-UNA in 2024, using data mining techniques.
Section II presents the theoretical framework of the study, focusing on educational data mining (EDM) and association rule techniques, as well as the most relevant algorithms. Section III describes the methodology used, including the design of the instrument and aspects of its application. Section IV analyses the results and discusses the main findings. Finally, section V summarizes the conclusions and suggests future research directions.
It is important to mention three specific points in the theoretical framework of this work. Below is an introduction to the concepts of educational data mining, association rules, and a discussion of their use in STEM work. 2.1.
According to [
The choice of methodological approach depends on the type of problem to be solved,
allowing multiple algorithms to be applied based on specific objectives. Various systematic
reviews, such as those presented in [
In the field of STEM and gender, studies such as [
This paper focuses on identifying relationships among different student characteristics using association rule algorithms. 2.2.
Association rule methods enable the discovery of frequent patterns between attributes in transactional datasets. These techniques facilitate the identification of relationships relevant to decision-making across various fields, including education.
In [
The strength of an association rule can be measured in terms of its support and
confidence. Support determines how often a rule applies to a given dataset, while confidence
determines how often the elements of Y appear in transactions containing X. The formal
definitions of these metrics are [
, ( → ) = , ( → ) = σ( ∪ ) σ( ∪ ) σ ( ) ( → ) = ( → ) ( )
In addition to support and confidence, another important metric is Lift, which is calculated as: of Y occurring.
>
1indicates a positive association, i.e., the occurrence of X increases the probability
1. Detecting frequent item sets, whose occurrence exceeds a minimum number of transactions (minimum support).
confidence level above the required threshold. 2. Obtaining the association rules associated with these frequent item sets, which have a The Apriori algorithm identifies frequent item sets in transaction databases and generates association rules. Its primary disadvantage is its high computational cost, as it requires multiple database scans. The FP-Growth algorithm is an efficient alternative to traditional association rule mining algorithms by compressing the dataset into a structure known as a Frequent Pattern Tree (FP), which allows the mining of frequent item sets without generating candidate sets
In education, various studies use rule mining to analyze student satisfaction, make
predictions in combination with other techniques, or identify common characteristics among
students [
For data collection, a structured survey instrument was designed, consisting of 33 questions distributed across six thematic dimensions. Table I presents each dimension along with the corresponding questions, their numbering within the instrument, and the type of response associated with each item.
To validate the instrument, the expert judgement technique was used [
Questions Can you tell us your age range? What is your gender? What degree are you studying? Do you consider that you have any specific learning needs or conditions that may influence your educational process? Select your year of admission Sociodemog
ráficos Employmen
t status Motivation
and expectation s 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
What type of secondary education did you receive?
Social Sciences, Technical Can you indicate your type of admission? If you entered the program through transfer, can you indicate your original program? What type of school did you come from? If you studied a technical secondary education program, was it in Informatics? How much time passed between finishing secondary school and starting this degree program? Did you take any programming courses before entering university? Have you participated in events and/or competitions related to technology, mathematics, or science? Can you tell us your area of origin? Can you tell us your city and department of origin? Can you tell us the city and county where you currently live? Can you tell us the educational level of the father/guardian Can you tell us the profession or postgraduate studies of your father/guardian? Can you tell us the educational level of your mother/guardian? Can you tell us your mother/guardian's profession or postgraduate studies? Do you have siblings who are professionals or students in the same field as you or in similar fields? Do you have significant family responsibilities that may require some of your time? Are you currently employed? Admission test, Direct admission, Transfer from another degree program, Transfer from another campus, Admission by Agreement. -Public, private, subsidised Yes, Other Less than 1 year, 1-2 years, 3-4 years, 5 or more years.
Yes, at the educational institution; Yes, self-taught; No Yes. Specify___ No.
Urban, Rural --Yes, No Yes, No, I prefer not to answer
Full-time, Part-time, No If you are working, is your job related to your career?
Yes, No What was the main reason you chose to study Computer Science? How long have you been interested in computer science or technology? What are your expectations regarding your career choice? Please indicate your level of interest in these areas.
Is there another area not mentioned that interests you and that you would like to indicate due to its relevance? Did anyone influence your decision to study this degree? Did you have any previous experience in programming or technology before entering university? Do you have access to a personal computer? Do you have access to the internet at home? -Yes, No Yes, No I am passionate about technology and computing, Good job and salary opportunities, Recommendation from family or friends, It is a career with a future, Other:___ Since childhood, Since secondary school, Recently, I'm not really interested To acquire advanced technical knowledge, To find a well-paid job, To start my own business or venture, To continue with postgraduate studies.
Very high, high, I don't know, low, not interested.
Yes, family members; Yes, friends; Yes, teachers; No, it was my own decision Yes, at school or college; Yes, self-taught (courses, tutorials, etc.); No, this is my first time Access to resources 32 33
The instrument was administered to new students in the 2024 cohort of the Informatics Engineering (IIN) and Bachelor of Science in Informatics (LCIk) programs, regardless of the admission method (test, transfer, agreement, or direct admission). Only data from female students was used for the analysis in this study.
Data collection was carried out using an electronic form sent by institutional email, accompanied by clear instructions. The form included information about the study's objectives, voluntary and anonymous participation, and the confidential use of data. Submission of the completed form was considered informed consent. In addition, first-year teachers collaborated in disseminating the instrument in their classrooms. The data were anonymized before analysis to protect the participants' identities.
Figure 1 shows the total female population of 54 students, 19 of whom are enrolled in the IIN program, while 35 are enrolled in the LCIk program. A total of 30 surveys were completed, 16 from the IIN program and 14 from the LCIk program. This represents an overall response rate of 55.56% of the total female intake population.
For processing the FP-Growth algorithm, we used the implementation of Python's mlxtend module2, which provides the function for calculating frequent item sets and the function for association rule mining. The survey data was pre-processed to generate a matrix format using One-Hot encoding, a technique that transforms categorical variables into binary format. In this matrix, each row represents a respondent, and each column corresponds to a specific combination of question and answer option; the value is True if that combination is present in the response.
In converting the instrument, each response option to categorical questions was transformed into an individual item following the format 'QuestionID_ResponseOption'. Open-ended questions were manually categorized, generating new categories for the most frequent or significant responses, and question 28, which corresponded to a Likert scale, was converted into a separate item. In the cases of questions 4 and 22, the option 'I prefer not to answer' was considered a valid category.
Given the sample size (N=30), the following parameters and criteria were established for the extraction and selection of rules: 2 MLXtend: https://rasbt.github.io/mlxtend/ ● ● ● ●
Minimum Support (min_support): This was set at 0.2, which means that a set of items must appear in at least six responses to be considered frequent. This value was defined empirically after preliminary tests, seeking to balance the number of patterns obtained with the manageability of the results in this exploratory study with a small sample. Minimum Confidence (min_confidence): This was set at 1.0, to identify only perfect implication relationships. This decision, appropriate to the exploratory approach and the small sample size (N=30), was made after iterative parameter adjustment to obtain a manageable and meaningful set of rules.
Lift metric: The lift metric was calculated for all generated rules. Rules with a lift greater than 1.2 are considered, as this indicates a positive association and that the items co-occur more than would be expected by chance.
Maximum item length: The length of the item sets was limited to two elements (max_len = 2) to facilitate the interpretation of the rules and ensure the manageability of the results. This decision responds to the exploratory approach of the study and seeks to ensure greater clarity in the analysis.
This section presents a general descriptive analysis of the responses obtained. 4.1.
17 You can tell us the educational level of the father/guardian Response 18 to 20
No
Public
Technical Less than 1 year 1 to 2 years
No Yes
No
Urban Primary Education Secondary Education
General (n=30) 24 22
IIN (n=16) 14 10
LCIk (n=14) 10 12 16 22 13 13 15 15 23 27 1 19 Can you tell us the educational level of your mother/guardian? 22 Do you have significant family responsibilities that may require some of your time? 23 Are you currently employed? 25 What was the main reason you chose to study Computer Science? 26 How long have you been interested in computing or technology? 27 What are your future expectations regarding your career choice? 28 Did anyone influence your decision to study this degree? 29 Did you have any previous experience in programming or technology before entering university? 2 4 2 3 5 3 1 16 18 12 26 10 10 19 14 9 13 11 16 7 1 3 0 3 4 8 1 0 10 6 14 8 5 11 5 7 6 6 7 6 1 1 2 0 1 8 2 1 8 6 5 8 9 2 7 5 9 12 4
There is a very similar proportion between those who have taken a programming course and those who have not. Most do not participate in STEM events and/or competitions and are not working when they start their careers.
Interest in computer science usually arises during secondary school or shortly before entering university. The main motivations for choosing this career are associated with the job opportunities and salaries offered by the sector.
All female students were admitted to their degree programs through an admission test. In general, their parents have a university education, and no influence from siblings was identified in their choice of degree program. All female students reported having access to a computer and the internet.
Figure 2 shows the main areas of interest. Among the most popular are Software Development, Video Games, Cybersecurity, Artificial Intelligence, and Machine Learning. The least interesting areas are Infrastructure and Networks, and IT Management. In the management area, there is a higher level of unfamiliarity than in other areas.
For the application of the FP-Growth algorithm, tests were carried out based on the combination of algorithm parameters indicated in the previous section. Twenty significant rules were identified. Table 3 shows the combination of questions and answers resulting in the rules found by the algorithm. For a better layout of the results, this data has been coded.
Have you taken any programming courses before entering university?_answer_YES, SELF-TAUGHT I13 I14 I15 I16 I17 I18 I19 I20 I21 I22 I23 I24 ● ● ●
Question and answer option Have you taken any programming courses before entering university?_answer_YES, AT AN EDUCATIONAL INSTITUTION Did anyone influence your decision to study this degree?_answer_NO, IT WAS MY OWN DECISION Can you tell us the educational level of your parent/guardian?_answer_UNIVERSITY Can you tell us your level of interest in these areas [IT Project Management]_answer_VERY HIGH Can you tell us your level of interest in these areas [Artificial Intelligence and Learning]_response_LOW Can you tell us how interested you are in these areas [Video Games]_response_NOT VERY INTERESTED Machine Can you tell us your age range?_response_18 TO 20 YEARS OLD What type of secondary education did you complete?_response_SOCIAL SCIENCES What type of secondary education did you complete?_response_TECHNICAL Do you have significant family responsibilities (caring for children, adults, etc.) that may require some of your time?_answer_NO Did you have any previous experience in programming or technology before entering university?_answer_NO, THIS IS MY FIRST TIME Did you have any previous experience in programming or technology before entering university?_response_YES, AT SCHOOL OR COLLEGE
I1 (event specification = nan): Given rule I1 → I9 (Rule No. 1 with conf=1.0), I1 is equivalent to I9 (Did not participate in events).
I2 (if Technical Secondary School was in Computer Science = YES): Given rule I2 → I21 (Rule No. 6 with conf=1.0), I2 is equivalent to I21 (Technical Secondary School). I3 (if Technical Secondary School was in Computer Science = nan): Given rule I20 ↔ I3 (Rule No. 3 and No. 4 with conf=1.0), I3 is equivalent to I20 (Secondary School in Social Sciences), or more precisely. The question about the specialization of the technical secondary school did not apply because the secondary school was in Social Sciences.
I1 I23 I20 I3 I7 I2 I24 I15 I10 I12 I17 I13
I9 I11 I3 I20 I14 I21 I21 I19 I19 I6 I4 I2 0,666667 0,333333 0,266667 0,266667
0,2 0,566667 0,533333 0,433333 0,233333 0,233333 0,266667 0,3 0,3 0,3 0,2 0,2 0,2 1 1 1 1 1 1 1 1 1,666667 1,578947 1,5 1,25 1,5 ● ● ● ● ● ● ● ●
Rule No. 2: Students who indicated that this is their first exposure to programming or technology (I23) always come from Asunción, the capital (I11). This pattern requires validation with future cohorts, as it could reflect a contextual relationship or sample bias. Rule No. 5: Students who expressed an interest in computer science or technology since childhood (I7) always indicated that they decided to study the degree on their own initiative, without external influences (I14). The high Lift suggests that a very early interest in the area encourages greater autonomy in choosing a degree.
Rule No 7, 12, 14: Those who had previous experience in programming or technology at their school (I24) took a technical secondary school (I21), and students who took a programming course at an educational institution before university (I13) also indicated that their technical secondary school was in informatics (I2). This indicates that programming experience acquired formally within secondary educational institutions is linked to technical secondary schools, mainly in informatics.
Rule No. 8: Students whose father/guardian has a university education (I15) are, in all cases, in the 18-20 age range (I19). In other words, the youngest group of entrants tends to have parents with a university education. This could be related to a family culture that promotes early access to higher education, although further exploration is needed to confirm this relationship.
Rule No. 10: Students who taught themselves programming before university (I12) always come from public schools (I6). This pattern could be related to the search for ways to supplement limited formal resources.
Rule No 15 and 16: Students whose main expectation is to acquire advanced technical knowledge (I5) always indicated that they had not participated in technological/scientific events before university (I9, and consequently I1). Those with a strong orientation towards advanced technical learning did not necessarily have a history of participation in events.
Rule No 18: This rule outlines a subgroup of students whose interest in technology was consolidated during their secondary school years (I8), but which does not extend strongly to the field of video games (I18). This suggests that interest in computer science is not necessarily linked to recreational use of technology, and that it can develop from more academic, professional or practical motivations.
Rule No 19 and 20: Students who expressed a 'Very High' interest in IT Project Management (I16) always indicated that they did NOT participate in events and/or competitions related to technology, mathematics or science (I9) before entering university. This association could serve to identify a student profile with a penchant for organization, strategic planning and leadership in technological contexts, rather than competitive or technical-practical activities. 4.3.
The main limitation is the small sample size (N=30). Although this sample allowed for a detailed exploratory analysis and the identification of 'perfect' associations within this specific group, the results are not statistically generalizable to the total universe of female entrants to the FP-UNA or other institutions.
The associations found, although they meet the strict metric criteria established (confidence=1.0, lift>1.2), should be considered as initial hypotheses that reflect strong patterns within the 2024 cohort.
Likewise, the sample size directly influenced the choice of algorithm parameters, requiring a restrictive approach (e.g., max_len=2, min_confidence=1.0) to obtain a manageable set of results, which may have left valid rules out of the analysis.
This paper constitutes a first exploratory approach to identifying association rules between various characteristics of female students who enrolled in 2024 in informatics programs at the FP-UNA. Based on the analysis carried out, the following findings stand out: ● ● ●
Regarding academic background: patterns linked to academic background were identified; for example, a perfect association between having completed a technical secondary school and previous experience in programming or technology, and a relationship between a secondary school in social sciences and non-participation in STEM events before university.
Regarding previous experiences and motivations: it was observed that certain previous experiences, such as self-taught programming, are associated in the sample with coming from public schools. Likewise, an early interest in technology (from childhood) was consistently linked to an autonomous decision to choose a career.
Regarding specific profiles: profiles were identified, such as female students who were interacting with technology for the first time but had developed an interest in secondary school, or those with a very high interest in IT Project Management who had not participated in previous STEM events.
As an initial exploratory study, these findings constitute a first approximation of the interrelationships between characteristics within this specific group of students. They provide guidance for future; more in-depth and far-reaching research focused on the characteristics and needs of female students in computer science programs at FP-UNA. It is hoped that the instrument will be applied to new cohorts of admissions, to increase the number of surveys. This would allow the rules found in this initial approach to be validated and, possibly, less restrictive support and confidence thresholds to be applied. Similarly, it would be important to include male students in the analysis to conduct comparative studies of profiles and gender-specific patterns of association. During the preparation of this work, the authors used Gemini and Grammarly in order to: Grammar and spelling check. After using these tools/services, the authors reviewed and edited the content as needed and takes full responsibility for the publication’s content.