Automation of Subject Analysis in Collection Evaluations
Every year we analyze an assortment of subject-based collections in order to make recommendations for program accreditation, enhancement purchases, and promotion. We typically start this process with a needs assessment and with analyzing collection-level trends. The final step of the evaluation process is to do a deep dive into each subject to examine a selection of raw and calculated measures (such as ILL requests, usage, recency, and the percent of Choice’s Outstanding Academic Titles list held) and assess whether the subject is strong, meets needs, or needs improvement. This process used to be completed manually, but with an average of 116 subjects per collection, this can be a very time-consuming, complex, and subjective process. In this presentation, we will discuss how we automated this process using a combination of SQL, Python, and regression to condense data about holdings, usage, ILL, and more into three scores for sufficiency, quality, and interest. We will also discuss how we use these scores to make recommendations that streamline the purchase process and enable targeted usage of our limited funding.