Things to do milestone 2

[AndresPitta]

Hey @pokrovskyy and @bradentam ,

Here is a checklist on the things to do for this milestone:

• [x] Script 1: Download/Save data script - Improve base on feedback given by the TA

Andrés:

• [x] Script 2: Data wrangling/prepreprocessing script. Arguments:
  • Raw data path
  • Cleaned data path
  • Test/train

Braden:

• [x] Script 3: EDA again. Arguments:
  • Cleaned data path
  • results.png

Serg:

• [x] Script 4: ML algorithm Arguments:
  • Cleaned data path
  • Results

Markdown Script - Report

• [x] introduction/justification of question - Braden
• [x] introduces the data set - Braden
• [x] presents the findings/results - Serg
• [x] interprets the findings/results in context of the question - Serg
• [x] Some critique of the analysis is also expected (limitations, assumptions, etc) - Serg
• [x] statement of future directions - Braden
• [x] 2 references (including the data source) and citing the packages - Andres

Proposal

• [x] implement feedback - Braden

Andres - update readme

• [x] usage (which order scripts are run in, what expected inputs are)
• [x] dependency flow chart

[AndresPitta]

Hey guys,

I just created a pull request with the data wrangling script.

these are the parameters it takes: Options:

--DATA_FILE_PATH= Path (including filename) to retrieve the csv file. [default: ../data/vehicles.csv] --TRAIN_FILE_PATH= Path (including filename) to print the train portion as a csv file. [default: ../data/vehicles_train.csv] --TEST_FILE_PATH= Path (including filename) to print the test portion as a csv file. [default: ../data/vehicles_test.csv] --TARGET= Name of the response variable to use. [default: price] --REMOVE_OUTLIERS= Logical value that takes YES as value if the outliers should be removed, NO otherwise. [default: YES] --TRAIN_SIZE= Decimal value for the train/test split. [default: 0.9]

What I did was removing prices over the 99th percentile, fixing the NAs for the categorical variables and splitting the train/test set. Every other variable seemed to be ok, based on the EDA.

Also, I haven't done the references for the libraries I used. Please, tell me what you think,