Key Factors That Predict Students’ Mathematics Competence in A College of Education in Hohoe, Volta Region of Ghana

This study investigates the demographic factors, including gender, age, program of study, and financial background, that predict students' mathematics competence in a College of Education in Hohoe, Ghana. A quantitative research predictive design was employed to examine the relationship between these demographic factors and mathematics competence. The study population consisted of 80 Science and Mathematics major students enrolled in a College of Education during the 2023 academic year. A simple random sampling technique was used to obtain 69 participants who successfully filled their questionnaire. The collected data were analyzed using several predictive models. Evaluation metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R²) were utilized to compare the performance of these models. Among the models compared, Boosting Regression demonstrated the best overall predictive performance. Random Forest Regression ranked as the second-best model, while RLR, KNNR, and NNR had poorer performance. The findings indicate that gender and program of study are consistently important factors in predicting students' mathematics competence. Additionally, age showed a weak positive association with mathematics competence, while financial status was inversely associated with performance. The results provide valuable insights for educators and policymakers, facilitating the development of targeted interventions to enhance students'


Introduction
Mathematics plays a crucial role in the advancement of science and technology, permeating numerous fields of study, such as physics, geology, engineering, biology, and medicine. In today's technologically evolving world, having a strong foundation in mathematics is considered a prerequisite for many careers and professions. For developing countries like Ghana, progress in industrial and technological development hinges on a workforce that is proficient in mathematics. Personal advancement in the current and future world also demands a solid understanding of mathematics and science. However, Wilmot and Otchey, (2012) opined that despite the increasing importance of mathematics, there is a noticeable decline in students' performance in the subject, particularly in tasks requiring a deep comprehension of mathematics. This trend is coupled with a significant decrease in students' interest in mathematics as they progress through high school. Mathematics is a crucial subject in many countries as it is considered essential for the development of the nation and for individuals to overcome challenges in daily life. Due to its significance, mathematics is given emphasis in schools at various levels globally (Capuno, et al., 2019). However, despite this emphasis, many students perform lower than the expected level in standardized national examinations, which has become a concern in many countries, including the Ghana. The underachievement in mathematics has been an issue in recent years, and it is important to address it. It is crucial to improve the students' knowledge and skills in mathematics to enable them to meet the demands of today's technological society As the world becomes increasingly focused on quantifiable data, possessing mathematical skills is essential for personal fulfilment and participation in education, society, and the job market in the 21st century. Karakolidis, et al., (2016), indicate that 36% of students fail to acquire basic mathematical skills in Greece while the average for European Union countries is much lower at 24.2%. Meanwhile, they stated that improving educational outcomes is crucial for increasing productivity and living standards, particularly in light of Greece's economic crisis and issues such as unemployment and educational inequality. Simply increasing educational spending is insufficient to enhance academic skills; a thorough examination of mathematics education and careful policy design are necessary. Nevertheless, there is a scarcity of research on mathematics achievement in this context, resulting in a lack of robust evidence regarding factors associated with students' mathematical performance. Kwame and Mary (20217) argued that every child has the right to access quality basic education. Education prepares children for social integration and economic freedom and can build significant analytical and social skills, enabling them to make good choices and pursue responsible lifestyles. As earlier expressed, Kwame and Mary emphasized that mathematics is a crucial subject that underpins scientific and technological knowledge and is therefore essential for the social and economic development of a nation. It is compulsory at both primary and secondary levels in Ghana and is a basic entry requirement for prestigious courses at the tertiary level. Mathematics enables individuals to understand the world around them and helps develop an analytic mind, leading to better organization of ideas and accurate expression of thoughts. It was noted that Mathematics enables one to make the invisible visible, thereby solving problems that would be impossible otherwise. They continue to state that Mathematics is widely used in various fields and covers a wide range of activities. Therefore, monitoring the progress of students in Mathematics at the Junior High School level is critical. The failure of students in Mathematics is worrying not only to educators but also to the students themselves. For example, the Ministry of Education Ghana in 2013 has shown that children in Ghana struggle not only with reading but also with Mathematics. Performance in Mathematics has lagged behind grade expectations, with the percentages of primary school pupils achieving proficiency in Mathematics falling below 20% (Ministry of Education, 2013). Similar results were reported in the 2016 (Ministry of Education, 2016). Additionally, over the last decade, there has been a remarkable drop in Mathematics performance among some Ghanaian students in national and international large-scale assessments such as the Basic School Certificate Examinations (BECE), West Africa Secondary School Certificate Examinations (WASSCE), and the Trends in International Mathematics and Science Study (TIMSS). A study by Wilmot and Otchey, (2012) aimed to explore the impact of Ghanaian junior high school students' and teachers' attitudes towards mathematics on students' performance in the subject. The study involved 400 students in their third year of junior high school, randomly selected from twenty schools. The findings revealed that both students' and teachers' attitudes towards mathematics significantly contributed to the students' achievement in the subject. The study suggested that junior high school mathematics teachers should receive training on how their attitudes towards the subject could affect students' performance, both during their pre-service and in-service education. In like manner, Karakolidis, et al., (2016) conduct a thorough analysis of the mathematical performance of 15-year-old students in Greece. Using data from the Programme for International Student Assessment 2012, the study employed a multilevel model to explore the factors associated with mathematics achievement at both the individual and school levels. The findings demonstrated that gender, pre-primary education, self-beliefs about mathematics, and individual and school-level socioeconomic status significantly predicted students' mathematical performance. Furthermore, the study highlighted the crucial role that school attendance played in shaping students' mathematical performance. The study suggested that a substantial portion of the variance in students' mathematical achievement could be accounted for by factors such as background characteristics, self-regulation, and school-level variables. Capuno, et al., (2019) also investigated the attitudes and study habits of 177 students in a public high school in the Philippines towards mathematics. The study found that the students had positive attitudes towards the value of mathematics, but were neutral in terms of their self-confidence, enjoyment, and motivation. There was a weak correlation between the value of mathematics and academic performance, and a weak positive correlation between attitudes and academic performance. The study recommended an enhancement plan in teaching mathematics to improve students' attitudes and study habits. As countries compete globally, education has become crucial to equip citizens with knowledge and skills for success. Research on educational effectiveness has increased in the past few decades, with scholars investigating factors that impact students' learning outcomes and social development across different levels within the school system. In Ghana, Butakor et al. (2017) conducted a study using multilevel modelling on the TIMSS 2011 mathematics data of students to examine variables that significantly contributed to students' performance in mathematics. The results indicated that Ghana's education system is similar to other systems where academic achievement is correlated with various student, classroom/teacher, and school characteristics. However, unlike other systems, the study found that school factors played a more significant role in the differences in students' mathematics achievement. The study identified several factors that may contribute to poor performance, including inadequate teacher preparation, emphasis on lower-level thinking skills, inconsistent use of homework, lack of engagement, lower educational aspirations of students, and gender inequality. These findings suggest the need for policy changes and improvements in teacher training to address the factors that may be limiting Ghanaian students' achievement in mathematics. Again, Kwame and Mary (2018) conducted a study to identify the factors that contribute to poor performance of students in mathematics in selected Basic Schools in a district in Accra, Ghana. The researchers used a descriptive research design and randomly selected 60 teachers for the study. The study found that the top school environmental factors that contributed to students' poor academic performance in mathematics were lack of supervision and monitoring of teachers by head teachers and circuit supervisors, larger class sizes, and a lack of teaching and learning materials. Untrained • Email: editor@ijfmr.com

IJFMR23045711
Volume 5, Issue 4, July-August 2023 4 teachers teaching the subject was found to be the most significant teacher factor contributing to poor performance. High levels of absenteeism among students and unruly student behavior were the most highly ranked student characteristics responsible for poor performance in mathematics. Regarding parental support, the top variables contributing to poor performance were parents not helping their children with homework, the inability of parents to provide essential instructional needs for mathematics studies, and parents' failure to approach the school to check on their children's progress in mathematics. However, the analysis did not find any statistical differences between the gender of teachers or their qualifications regarding the perceived causes of poor performance in mathematics. The researchers recommended that metropolitan, directorate, and circuit supervisors intensify their supervision and provision of learning materials in schools. This study aims to investigate the key factors that predict students' mathematics performance in a college of education in Hohoe, Ghana. Based on the findings from existing literature, both student and teacher attitudes towards mathematics have a significant impact on students' achievement. Additionally, gender, pre-primary education, self-beliefs about mathematics, and individual and school-level socioeconomic status have been identified as significant predictors of students' mathematical performance. The literature also highlights factors including inadequate teacher preparation, emphasis on lower-level thinking skills, inconsistent use of homework, lack of engagement, lower educational aspirations of students, and gender inequality. The justification for this study is based on the findings from the literature review, which indicate that there are several factors that predict students' mathematics performance. However, there is a gap in the literature on the specific factors that predict mathematics performance among students in a college of education in Hohoe, Ghana. Therefore, this study aims to investigate the key factors that predict students' mathematics performance in this specific context, including gender, age, program of study, and financial status. In this study, the key factors to be investigated include gender, age, program of study, and financial status. The identification of these factors and their influence on students' mathematics performance will provide insights that can inform policy and practice in the college of education and beyond.

Research objectives
The primary objective of this study is to investigate the demographic factors, including gender, age, program of study, and financial background, that predict students' mathematics competence in a College of Education in Hohoe, Ghana. Specifically, the study aims to: 1. Examine the relationship between gender and students' mathematics competence. 2. Investigate the influence of age on students' mathematics competence.
3. Analyze the association of program of study on students' mathematics competence. 4. Explore the relationship between students' financial background and their mathematics competence.

Literature Review Demographic factors and Mathematics Competence
Belhu (2017) investigate the variables affecting math instruction and learning among students. A descriptive survey method with qualitative analysis was used in the study. investigating the variables that affect mathematics teaching and learning was the main purpose of the study. Data was gathered using a variety of tools, including interviews, focus group discussions, and questionnaires. In all, 80 students along with sixteen teachers, responded to the questionnaires. The results showed that individual traits, instructional tactics and methods, and demographic factors all have an impact on how mathematics is • Email: editor@ijfmr.com IJFMR23045711 Volume 5, Issue 4, July-August 2023 5 taught and learned. The most significant of these were found to be individual and demographic characteristics, followed by instructional tactics and methodologies. The respondents frequently encountered difficulties with language, finances, general knowledge, and confidence in their grasp of mathematical ideas. The investigation also discovered that the university lacked the tools and resources necessary to teach and learn mathematics. However, it was acknowledged that adequate resources and infrastructure were necessary for effective mathematics teaching and learning. Based on the study, it was advised that students possess the requisite background knowledge and mathematics skills and build their mathematical confidence. Papadakis et al. (2016) examines and contrasts the effects of computer and tablet use on young children's development of mathematical proficiency. The study was carried out for 14-weeks intervention to administer the survey, with one experimental and one control group. Both groups received mathematical instruction in accordance with the Greek early childhood curriculum while using the same educational software, which, depending on the group, was either operating on PCs or tablets. The Test of Early Mathematics Ability (TEMA-3) was used to assess children's mathematics performance. The sample included 256 Greek children. The findings demonstrated that teaching with tablets has greatly contributed to the development of children's arithmetic abilities to a larger extent than teaching with computers. Furthermore, it noted that age or gender did not make a difference in how well children developed their mathematical skills. Global socioeconomic achievement disparities in mathematics have grown significantly over the past few decades, according to international studies. The disparities in the personal and familial qualities of the pupils may be a contributing factor in these achievement gaps. If educational systems offer various opportunities to learn for students from rich vs disadvantaged environments, they might also be attributed to the education itself. Previous studies on the joint link between socioeconomic status, academic success, and opportunities to learn yielded conflicting findings. By reanalyzing PISA data, the primary goal of the study is to determine if education genuinely contributes to socioeconomic gap in accomplishment. The study pays close attention to the construct validity of the opportunity to learn measure in PISA that has been applied in other studies. The findings suggest that the PISA's opportunity to learn indicators have two hidden dimensions: an unbiased opportunity to learn dimension and a self-concept component. When the impact of students' self-concept was taken into account, opportunities to learn role in mediating the association between social background and mathematics achievement was only marginally significant. The findings implied that there is a construct validity issue with the earlier research indicating that education reinforces social differences in mathematics achievement.

Gender, Age, program of study and financial differences in mathematics performance
Osadebe and Oghomena (2018) evaluated the sociodemographic factors influencing students' mathematics performance. The goal of the study is to determine how students' performance in mathematics on the Senior Secondary Certificate Examination relates to gender, location, socioeconomic position, and other factors. The study employed an ex-post facto research design. 15,170 SS3 people make up the study's population. Using stratified random sampling and simple random sampling techniques, a sample of 759 students was randomly chosen from the entire population. A 40-item multiple-choice senior secondary mathematics achievement test (SSMAT) was the tool employed for the investigation. The analysis made use of multiple regressions. The study found that student's performance in mathematics on the senior secondary certificate examination was influenced by gender and socioeconomic class. Another study compared the arithmetic abilities of first graders from Taiwan, Russia, and the United States based on gender. Children (N=250, Mean age=7 years 2 months) expressed their solutions to simple (single-digit) and complex (mixed-and double-digit) addition problems (Shen et al., 2016). The study indicates that there were gender disparities in techniques for straightforward issues that differed among nations, but there were no gender differences in accuracy. Students from the United States and Russia used different strategies while solving complicated tasks, with gender differences mediating differences in accuracy. In contrast, there were no gender differences in tactics or accuracy among Taiwanese students. The pattern of findings indicates that gender inequalities in mathematics may be influenced by the school environment. Also, Mejía-Rodríguez et al. (2021) performed series of mean comparisons and regression analyses using data from the TIMSS 2015 assessment of fourth-grade students in 32 countries to ascertain the following: (a) the gender gap in students' self-concept in mathematics; (b) the relationship between student achievement, student gender, and parental characteristics (early numeracy activities, attitudes, expectations, and education) and students' self-concept; and (c) the impact of achievement and parental characteristics. The findings of this study show that there are considerable gender disparities in students' self-concepts in mathematics across most nations, with males often benefiting from these differences as early as the fourth grade. When the data analysis takes into account the effects of student achievement and parental involvement, the discrepancies mostly hold true. Similarly, Ghasemi and Burley (2019) argued it has been a persistent problem to figure out why women are consistently underrepresented in STEM disciplines, with math ability being a recurring element of the conundrum. This study examines cross-national gender variations in math-related affect, namely liking mathematics, having confidence in mathematics, and appreciating mathematics, using data from TIMSS tests. To look for any variations in these gender-related emotional traits, we compared fourth and eighth graders. The results imply that boys and girls are similar in spite of variation and some changes to the amount and direction of gender differences in math affect. In fact, the researchers discovered that students in some nations with lower adult gender gaps have bigger gender differences in affective factors related to mathematics.

Methods and Materials
This study utilized a quantitative research predictive design to examine the relationship between demographic factors and students' mathematics competence in a College of Education in Hohoe, Ghana. The quantitative approach allows for the collection of numerical data, enabling statistical analysis to determine the predictive power of various demographic variables. The study population consist of 80 Science, ICT and Mathematics (SICTM) major students enrolled (2023 academic year) in a College of Education in Hohoe, Ghana. The simple random sampling technique was employed to ensure representation from the three programs of study and demographic groups. The sample size was determined based on Krejcie and Morgan (1970) sample size table to ensure adequate representation and meaningful results. The primary data collection instrument (survey questionnaire) was used. The questionnaire was designed to gather information on the demographic variables of interest, including gender, age, program of study, and financial background. Additionally, the questionnaire included items related to students' mathematics competence, such as self-assessment of mathematical skills and perceptions of mathematical abilities. The collected data was analyzed using predictive models (Regularized Linear Regression, Random Forest Regression, Neural Network Regression and Boosting Regression). The study ensures that all ethical guidelines for research involving human participants are followed, including obtaining informed consent, ensuring confidentiality and anonymity, and addressing any potential risks or concerns. The validity and reliability of the survey questionnaire was done using appropriate statistical techniques (e.g., Cronbach's alpha for internal consistency). The study was limited in relation to sample size constraints and generalizability of the findings.   The Neural Network regression model (NNRM) result suggests that the four predictor variables (gender, age, program of study, and financial status) have different associations with the target variable (student performance), and that the model is performing reasonably well in making predictions. The validation MSE of 1.442 and test MSE of 0.607 suggest that the model is performing relatively well in making predictions on the data. The NNRM suggests that gender, age, program of study, and financial status are all important predictors of student performance, and that the model may be useful in predicting performance for new students.  . In this specific case, the model was trained on a dataset containing 69 students, and was then tested on a validation set of 18 students and a test set of 21 students. The hyperparameters used were:

Result and Findings
• Nearest neighbors = 3: This means that for each prediction, the model looks at the 3 closest neighbors in the training set and uses their average target value as the predicted value for the new data point. • Weights = rectangular: This means that all the nearest neighbors are given the same weight when making the prediction. • Distance = Euclidean: This is the distance metric used to measure the distance between data points in the input space. The validation and test mean squared errors (MSE) are used to evaluate the model's performance. The validation MSE was 0.641, which means that on average, the model's predictions on the validation set were off by 0.641 units (squared). The test MSE was 0.796, which means that on average, the model's predictions on the test set were off by 0.796 units (squared). The study's findings underscore the importance of considering gender, program of study, and age as influential factors in teaching and supporting mathematics education. It is crucial for educators and policymakers to address these factors to ensure equitable access and opportunities for all students to succeed in mathematics.

Variable Trace Plot Total Increase in Node Purity
Logistic Sigmoid Activation Function Relative Influence Plot Figure 1: Pictorial representation the model influence performance

Conclusion
Based on the information provided, five different predictive models have been used to predict student performance based on four characteristics: gender, age, program of study, and financial status. These models are Ridge Linear Regression, Random Forest Regression, Neural Network Regression, K-Nearest Neighbors Regression, and Boosting Regression. In conclusion, the results indicate that each model can produce accurate predictions based on the given dataset, but some models may perform better than others based on the specific characteristics being analyzed. For instance, Ridge Linear Regression uses regularization to prevent overfitting and has performed well when predicting student performance based on gender, age, program of study, and financial status. In contrast, Random Forest Regression, Neural Network Regression, and K-Nearest Neighbors Regression use different techniques to build complex models that can capture complex interactions between the input features, which have produced accurate predictions in previous studies. Boosting Regression is another model that has been used to predict student performance, and it assigns weights to each of the four characteristics, indicating their importance in predicting student performance. In this model, gender and program of study are the most important factors in predicting student performance, while age and financial status have less impact.

Recommendation
In view of the result, findings and conclusion, it is recommended that; 1. Colleges of Education management should be guided against student gender and choice of program of study when admitting new students so that student. This can be done during fresh student orientation before they finally settled on their choice of study. 2. Parents and caretakers of prospective college student should be monitored and during their pre-tertiary education in order to counsel them against their studies in college. This should be done considering gender, strength and weaknesses subject areas well as the ages. 3. Although, financial status of student was the least significant influence on student performance.
College student caretakers and institutional management should make conscious effort to support students have financial stability while on campus. 4. Ultimately, the predictive models can help educators and administrators make data-driven decisions to support student success and improve academic outcomes.

Limitation and Suggestion for Further Studies
The RLR though was good in predicting the key variables is worth nothing that there was an issue of overfitting. However, it is important to consider the limitations of the data and the modeling approach when interpreting and applying these results. The current model only includes four predictor variables. It may be useful to consider additional variables that may be relevant to student performance, such as prior academic achievement, motivation, or social support. However, adding more variables may also increase the risk of overfitting, so it is important to carefully evaluate the trade-off between model complexity and predictive accuracy. 1. Each model has its strengths and weaknesses, and the choice of model depends on the specific characteristics being analyzed, the size of the dataset, and the performance metrics used to evaluate the models. 2. The models can be further improved by incorporating additional data and refining the algorithms used to train the models.