Individual Assignment Tahoe Healthcare Systems Case Analysis

联系我们： 手动添加方式: 微信>添加朋友>企业微信联系人>13262280223 或者 QQ: 1483266981

Data Analytics and Visualisation
Jiankun Sun
Imperial Business School
Individual Assignment
Tahoe Healthcare Systems Case Analysis
1 Overview
In this assignment you will act as a data analyst advising the management of Tahoe Health
care Systems. Read the case (Tahoe healthcare case.pdf) carefully before you begin. The
data is provided in Tahoe Healthcare Data . csv, with a training/test split in train . csv and
test . csv.
Your task is to prepare a business analytics report for Tahoe’s senior management team to
answer their central question:
Should Tahoe deploy the Care Tracker program, and if so, for which patients
Your report should use the data to develop a clear recommendation. It should combine de
scriptive analytics, predictive analytics, economic reasoning, and data storytelling. The reader
should be able to understand what you learned from the data, how your analysis supports the
decision, and what Tahoe should do next.
Note: Treat the dataset as representative of what will happen in a given year if nothing is done
to reduce the readmission rate.
2 Report Structure
Organise your report as follows:
2.1 Executive Framing
Begin by explaining the decision Tahoe faces. Define the business problem, explain why it
matters, and state the key analytical questions your report addresses.
2.2 Descriptive Analytics
Use descriptive analytics and visualisation methods to explore the data and develop a brief
understanding of patient characteristics and readmissions. This section should help the reader
see the baseline situation, the relevant patient characteristics and readmission information, and
any patterns that may matter for your later modelling and decisions.
You are free to choose the analyses and visualisations, but focus on those that are useful for the
case rather than showing everything. Conclude this section with a summary of the key insights
you have gained as a data explorer — in particular, the findings that can inform your later
predictive modelling choices and managerial interpretations.
Page 1 of 3
2.3 Business and Economic Case
Develop the economic logic of the CareTracker decision. This section should explicitly address
the following:
1. Under the status quo, how much will Tahoe lose in Medicare reimbursements due to 30-day
AMI readmissions
2. If CareTracker is implemented for all AMI patients, what is the net financial impact relative
to the status quo
3. If Tahoe had perfect foresight about which patients would be readmitted, what is the max
imum possible saving from targeting CareTracker
Use these calculations to explain why targeted intervention may be more attractive than uni
versal implementation. The goal is not only to calculate costs and savings, but also to explain
the economics of the decision.
2.4 Predictive Analytics
Develop a classification approach for 30-day readmission risk. The target variable is readmit30 .
2.4.1 Simple Severity-Threshold Benchmark
Start by using the severity score as a simple benchmark classifier. Consider rules of the form:

reat30 =
1 , if severity score S > S ,
0 , otherwise.
Using the full dataset (Tahoe Healthcare Data.csv), compute the cost savings relative to the
status quo for threshold values S from 25 to 100 in increments of 1 . Create a graph of estimated
cost savings as a function of S . Identify the best threshold according to cost savings and explain
what this benchmark teaches you about risk targeting.
2.4.2 Developing and Selecting a Better Model
Develop better predictive models for readmissions. Experiment with multiple modelling ap
proaches and/or variable specifications to improve performance. Train your models on train . csv
and evaluate them on test . csv.
Your main report should present the model you recommend. The report should also explain
how the model performs on the test data;
how much economic value it creates relative to the status quo and relative to simpler alter
natives;
how you evaluate model performance and why you select the final model over other models
(i.e., how you select the decision rule);
2
what managerial or clinical insights can be learned from the model;
any limitations, risks, or implementation concerns.
Detailed comparisons of all models, variables, parameter settings, and performance results
should be placed in the appendix.
2.5 Recommendation
End with a clear recommendation for Tahoe’s senior management team. State whether Care
Tracker should be implemented universally, targeted to selected patients, or not implemented.
Explain the evidence behind your recommendation, the expected financial implications, and
any important limitations or next steps.
3 Submission Requirements
Please submit the following files:
1. A main report (PDF) with an appendix. The main body of the report should be no more
than three pages, excluding the appendix. The appendix may include supporting analy
ses, additional figures, model comparisons, parameter choices, sensitivity checks, or other
technical details, and has no page limit.
2. Your Excel workbook, or other code files, that can reproduce your analysis.

Individual Assignment
Tahoe Healthcare Systems Case Analysis最先出现在KJESSAY历史案例。