Luiza Antonie

Tracking people over time in 19th century Canada for longitudinal analysis

Linking multiple databases to create longitudinal data is an important research problem with multiple applications. Longitudinal data allows analysts to perform studies that would be unfeasible otherwise. We have linked historical census databases to create longitudinal data that allow tracking people over time. These longitudinal data have already been used by social scientists and historians to investigate historical trends and to address questions about society, history and economy, and this comparative, systematic research would not be possible without the linked data. The goal of the linking is to identify the same person in multiple census collections. Data imprecision in historical census data and the lack of unique personal identifiers make this task a challenging one. In this paper we design and employ a record linkage system that incorporates a supervised learning module for classifying pairs of records as matches and non-matches. We show that our system performs large scale linkage producing high quality links and generating sufficient longitudinal data to allow meaningful social science studies. We demonstrate the impact of the longitudinal data through a study of the economic changes in 19th century Canada.

Negative Association Rules

Mining association rules associates events that took place together. In market basket analysis, these discovered rules associate items purchased together. Items that are not part of a transaction are not considered. In other words, typical association rules do not take into account items that are part of the domain but that are not together part of a transaction. Association rules are based on frequencies and count the transactions where items occur together. However, counting absences of items is prohibitive if the number of possible items is very large, which is typically the case. Nonetheless, knowing the relationship between the absence of an item and the presence of another can be very important in some applications. These rules are called negative association rules. We review current approaches for mining negative association rules and we discuss limitations and future research directions.

Classifying microarray data with association rules

In this paper we investigate a method for classifying microarray data using association rules. Associative classifiers, classification systems based on association rules, show good performance level while being easy to read and understand. This feature is especially attractive for biological data where experts can read and validate the association rules. Relevant features are selected using Support Vector Machines with Recursive Feature Elimination. These features are discretized according to their relative expression levels (upregulated, downregulated or no change) and then they are used to build an associative classifier. The proposed combination proves highly accurate for the studied microarray data collection. In addition the classification rules discovered and employed in the classification process prove to be biologically relevant.

Dancing with Dirty Data: Problems in the Extraction of Life-Course Evidence from Historical Censuses

This chapter builds on a recent use of SVM classification to create linked sets of Canadian 1871 and 1881 census records. The census data are imprecise and have limited granularity; many records share identical detail. In spite of these challenges, the SVM generates life-course information for large numbers of individuals with a low (3 %) false positive error rate. However, there is a higher incidence of error among apparent migrants when the true rate of migration is small. The linked data are broadly representative of the population with some underrepresentation of illiterates, young adults (especially unmarried women), older people (especially men), and married people of French origin. The new longitudinal data are of considerable research value but users must take into account these weaknesses.

Redundancy reduction: does it help associative classifiers?

The number of classification rules discovered in associative classification is typically quite large. In addition, these rules contain redundant information since classification rules are obtained from mined frequent itemsets and the latter are known to be repetitive. In this paper we investigate through an empirical study the performance of associative classifiers when the classification rules are generated from frequent, closed and maximal itemsets. We show that maximal itemsets substantially reduce the number of classification rules without jeopardizing the accuracy of the classifier. Our extensive analysis demonstrates that the performance remains stable and even improves in some cases. Our analysis using cost curves also provides recommendations on when it is appropriate to remove redundancy in frequent itemsets.

Analyzing the Gender Wage Gap in Ontario's Public Sector

In this paper, we analyze the gender wage gap in Ontario?s public sector. Our analysis is based on the salaries of high earners in the public sector. Although these salaries are publicly available from Ontario?s Sunshine List, a key attribute is missing from the public data, the gender variable. We propose a 2-stage model to predict the gender based on the person?s first name, and we augment the data with the new variable. With the new database created, we analyze, present and discuss results for the gender wage gap in Ontario. The findings of this research are being used by Ontario?s provincial government to reassess and change current policies for pay equity.

Comparing Classifiers in Historical Census Linkage

Linking multiple data collections to create longitudinal data is an important research problem with multiple applications. Longitudinal data allows analysts to perform studies that would be unfeasible otherwise. In our research we are interested in linking historical census collections to create longitudinal data that would allow tracking people overtime. The goal of the linking is to identify the same person in multiple census collections. A classification system is employed to make the decision if two people are the same or not, based on their characteristics. In this paper we present an empirical study where we explore the use of three different classifiers in a record linkage system and we evaluate their performance.

Biologically relevant association rules for classification of microarray data

In this paper we investigate a method for classifying microarray data using association rules. Associative classifiers, classification systems based on association rules, show good performance level while being easy to read and understand. This feature is especially attractive for biological data where experts can read and validate the association rules. Relevant features are selected using Support Vector Machines with Recursive Feature Elimination. These features are discretized according to their relative expression levels (upregulated, downregulated or no change) and then they are used to build an associative classifier. The proposed combination proves highly accurate for the studied microarray data collection. In addition, the classification rules discovered and employed in the classification process prove to be biologically relevant.

Gender Wage Gap in the University Sector: A Case Study of All Universities in Ontario, Canada.

By analyzing salary data from the Ontario Sunshine List for the University Sector and combining it with productivity characteristics for research and teaching, we extend our understanding of the variables that contribute to the gender wage gap in Academia. Longitudinal analysis confirms that employees labelled as female are less represented in administration roles and full faculty positions. While the gender imbalance on the list is getting less extreme, with the proportion of women on the Sunshine List increasing from 11% in 1997 to about 40% nowadays, this increase in female representation is more likely to occur at incomes close to the access threshold of

Population Analysis of the Settlement Movement in Western Canada

100,000. While women do not achieve wage parity even when sorted by faculty position, within each academic rank the gender wage gap is smaller than the overall wage gap, which further confirms that, even in the ivory tower, men select into more lucrative positions than women. Controlling for productivity measures for research with h-index and for teaching with overall Rate My Professors (RMP) shows a modest effect of these productivity measures on wage formation and no effect on the wage gaps.