marthahunter / teis-location-analysis Goto Github PK

In this project, I worked with a group of my peers in my Data Analytics cohort. We analyzed one year's worth of Tennessee Early Intervention System (TEIS) data to provide insights regarding evaluation scores across regions and subregions of Tennessee. Several groups split up and answered different questions. Then we put together this presentation and presented to the TEIS representative who introduced us to the data and provided us with her questions.

This was my group's question at hand:

7. Is there any notable pattern of scoring by region? Note: TEIS contracts with three agencies for evaluations (one per grand region) as follows:

• East TN, First TN and Southeast
• Greater Nashville, Upper Cumberland, and South Central
• Northwest, Southwest, and Memphis Delta

We utilized Python/Jupyter notebooks and Seaborn visualizations to understand and demonstrate different regional and subregional trends within the dataset TEIS provided. Below, you can see a couple graphs we generated. Other analysis included average percentile rank by region, average developmental quotient by region, and heat maps to present specific subdomain evaluations by age and explain how scores varied among regions.

I also served on the Editorial Team (also known as the Project Management Team) to set style guidelines and other parameters. This helped ensure a uniform voice and look to our class presentation. The image below is a list of guidelines I helped generate and communicate out to the group, all on an expedited timeline. These guidelines included custom color palette dictionaries, font pairings, Seaborn themes, capitalization guidelines for titles and axis labels, and more. Serving on the Editorial Team was a fun and challenging way to transfer skills from my experience in marketing to a data analytics project.

The Tennessee Early Intervention System (TEIS) is a voluntary program that offers therapy and other services to families of infants and young children with developmental delays or disabilities. Services are provided at no cost to families. TEIS is critically important to supporting the development of Tennessee children with disabilities and developmental delays as they prepare for school.

In this project, you are going to be working with data gathered from the Battelle Developmental Inventory, Third Edition (BDI-3) Eligibility Evaluation, an early childhood instrument based on the concept of developmental milestones. The BDI-3 is used by TEIS for determining eligibility and outcomes. This assessment tool has been in use for less than a year now, so the purpose of your work is to analyze the scores to see if there appear to be any weaknesses in this tool in any of the assessed domains or subdomains.

The BDI-3 is divided into 5 domains, each of which has 2 or 3 subdomains. On each domain and subdomain, a child receives several different scores. An example score report is available in the BDI3_ScoreReport.pdf file. See page 3 of this file for a detailed description of each score. Note that the BDI-3 is norm-based, meaning that the scores are generated by the child's performance based on the performance of same-aged peers on the same items.

Some notes about the data:

• The same instrument is used for determining eligibility and outcomes. The type of evaluation can be determined from the "Program Label" column, which will indicate either an eligibility evaluation, an annual evaluation, or a birthday/milestone or exit evaluation.
• Some of the rows are duplicated, so be sure to check that in your initial data inspection and cleaning. In addition, some children will have multiple rows in the dataset. A child is evaluated at initial eligibility, annually, and at their 3rd birthday or exit under certain circumstances. The evaluation may be repeated every 4 months (e.g. a child was ineligible and re-referred).
• There are many records that are missing ID numbers. These records have been given a unique identifier which starts with "SP" so that they can be identified.
• When reading the data in, you may notice that some of the values in the RDI columns are incorrectly converted to dates. It is likely that this happens whenever the numerator of the fraction is 12 or less.

Project Objectives:

1. In which domains (and sub-domains) are children performing highest and lowest?

2. What is the pattern of children's scores? Do individual children tend to perform similarly across all domains or are there differences from domain to domain? Are these patterns similar for all children?

3. Is there a pattern of significant differences in children's scores in the sub-domains within a domain? For example, a gap between a child's expressive (talking) and receptive (understanding) language scores may indicate that additional evaluation is needed for that child's eligibility. A gap between these sub-domains for all or most children may indicate the BDI-3 tool may lack sensitivity in the language domain.

4. Does the child's age seem to impact their scoring?

5. Eligibility is currently based on domains. A child would be considered eligible based on a 40% delay in one area or 25% delay in two areas, as determined by Development Quotient (DQ) scores. A DQ of 70 or less indicates a 40% delay, and a DQ between 71 and 78 indicates a 25% delay. If TEIS added an option for eligibility based on the total score on the BDI-3 total score (DQ score of 70 or less), are there any children who would have qualified that did not qualify based on the current method?

6. Is there any difference in scoring noted based on evaluation type?

7. Is there any notable pattern of scoring by region? Note: TEIS contracts with three agencies for evaluations (one per grand region) as follows:

• East TN, First TN and Southeast
• Greater Nashville, Upper Cumberland, and South Central
• Northwest, Southwest, and Memphis Delta

8. Is there any notable pattern of scoring by evaluator?

Stretch Goal:

• Investigate the records that are missing ID values. Have they gotten better or worse over time? Are there repeat offenders?
• As noted above, this evaluation may be repeated every 4 months. How often did it occur that a child was assessed multiple times in a time period of less than 120 days. Has this gotten better or worse over time?