代写MSBA7002、代做Statistics、代写R编程语言、R设计代做

日期：2019-11-27 08:59

Predict the Bankruptcy Situation of Polish Companies

MSBA7002: Business Statistics

Nov 13, 2019

Due Date: 11:55pm Dec 1, 2019

Projective Deliverables

(1) Power point or pdf file, containing clear steps about model selection and five

interesting visualizations that can help to answer analytical problems;

(2) Rmarkdown file that embeds R code within the analysis narrative, or

(3) Python code with comments written in Jupyter notebook.

(please make sure that by running the code, all the results reported can be

reproduced).

Content

Bankrupt file

is about bankruptcy prediction of Polish companies. The data contains financial rates

from 2nd year of the forecasting period and corresponding class label that indicates

bankruptcy status after 4 years. The bankrupt companies were analysed in the period

2000-2012, while the still operating companies were evaluated from 2007 to 2013. In

this task, you need to use 64 features from the financial reports to predict which

companies will go bankrupt in the next four years.

Lookup file

contains the text description of all variable codes in the data file.

Training and testing datasets

Training data: containing 6000 rows of data;

Testing data: containing 3000 rows of data.

Tasks

1. Data Pre-processing

This dataset contains plenty of missing values. You need to report how you handle

these missing values.

2. Model Selection

Report how you build your model or model ensemble with suitable criterion.

3. Visualizations

Report 5 most interesting visualizations that can help to answer analytical problems.

For example, are there any predictors have some high correlations, which two

dimensions can provide a good classification performance, etc.

4. Classification

Based on the features available, develop a model that predicts the bankruptcy

situation of the companies. The classification results will be evaluated in Kaggle

automatically.

The evaluation metric – F score

with TP, FP and FN being the numbers of true positive,

false positive, and false negative, respectively.

5. Report results

The slides should include at least three parts – visualization, methodology, and results.

The methodology section should be precise and can justify your decisions, for

example, how you choose hyperparameters, why you prefer a particular method over

the others?

The codes need to demonstrate that the classification results are reproducible, and the

adopted method is consistent with the one introduced in the submission file.

Notice

1. Project deadline: 11:55 pm, Dec 1, 2019.

2. You can use either R or Python for the classification task.

3. The composition of marks is given in the table below

Total Score Criterion

Visualization 5 Innovation

Aesthetic

Information complexity

Insightfulness of conclusion

Classification 1

Analysis

Presentation

10 Clear idea about model selection

Good interpretation about model

Well-structured report