/JdMS

Building a segmentation model using clustering

Primary LanguageJupyter Notebook

JdMS

Edit Friday, June 14th 2019

Our files:

  1. DataCleaning.ipynb: creating the cleaned dataset, quality assurance, and modifying the data frame. Export the clean data as:
  • export the data (long format - transaction table) as cleaned_data.parquet
  • export the data (wide format - customer table) as customer_data.csv

  1. TDataVisualization.ipynb: creating graphs to show relationships/trends not related to the model. Uses the transaction table.

  2. CDataVisualization.ipynb: creating graphs to show relationships/trends not related to the model. Uses the customer table.

  3. Kmeans.ipynb: this is the segmentation model that utilizes k-means to find our clusters.

  4. MeanShift.ipynb: this is the segmentation model that utilizes mean shift to find our clusters.

  5. DBSCAN.ipynb: this is the segmentation model that utilizes DBSCAN to find our clusters.

  6. GowerDistance.ipynb: this is the file that calculates gower's distance for the customer table.

  7. MeanShift.ipynb: this is the model running with the mean shift algorithm.

  8. PCA.ipynb: runs the cleaned customer table through Principal component analysis, reduces the dimensions of the data so it can run smoother.

  9. UMAPS.ipynb: runs the data through UMAPS.

Over the course of the project we will add notes about the files of the project and what we are doing.