/bulldozer-price-prediction

Machine Learning project to predict the auction price of bulldozers.

Primary LanguageJupyter Notebook

Bulldozer Price Prediction

In this project, we'll use a Bulldozer Price dataset taken from Kaggle, to create a regression machine learning model that will predict the auction price of a bulldozer, given different factors, such as different machine configurations and year of manufacturing.

Take a look at the finished notebook.

Getting the data

You can download the full data from the link mentioned above. Or you can download this zip with the used files and the correct folder structure for the Notebook.

Installation

First, clone this repository to your machine:

git clone https://github.com/JorgePasco1/bulldozer-price-prediction

Using conda

This project uses conda as a environment/package manager. To easily run this project in your machine, you should have ananconda or miniconda installed (You might want to refer to Conda documentation).

Then, create the environment folder with the dependencies within the project directory:

conda env create --prefix ./env -f ./environment.yml

Activate the environment, with:

conda activate ./env

And finally, if dependencies are correctly installed, you should be able to run:

jupyter notebook

Troubleshooting

When creating the enviroment, you may encounter ResolvePackageNotFound problem:

ResolvePackageNotFound:
  - dependencies not found

To solve the problem, move the problematic dependecies under pip section in the .yml file, like:

dependencies:
  - appnope=0.1.0=py38_0
  - attrs=19.3.0=py_0
  - backcall=0.1.0=py38_0
  - blas=1.0=mkl
  - bleach=3.1.4=py_0
  - ca-certificates=2020.6.24=0
  - certifi=2020.6.20=py38_0
  - cycler=0.10.0=py38_0
  - dbus=1.13.14=h517e14e_0
  - decorator=4.4.2=py_0
  - defusedxml=0.6.0=py_0
  ** all dependencies, except those not found **
  - pip:
    - jedi=0.17.0=py38_0
    - jinja2=2.11.2=py_0
    - joblib=0.15.1=py_0
    - jpeg=9b=he5867d9_2
    - jsonschema=3.2.0=py38_0
    - jupyter=1.0.0=py38_7
    - jupyter_client=6.1.3=py_0
    - jupyter_console=6.1.0=py_0
    - jupyter_core=4.6.3=py38_0
    - kiwisolver=1.2.0=py38h04f5b5a_0
    - libcxx=10.0.0=1
    - libedit=3.1.20181209=hb402a30_0
    - libffi=3.3=h0a44026_1
    - libgfortran=3.0.1=h93005f0_2
    - libiconv=1.16=h1de35cc_0
    - libpng=1.6.37=ha441bb4_0
    *** All dependencies not found ***

No Conda

Alternatively, you can use your preferred method to install the required dependencies (jupyter, NumPy, Pandas, Matplotlib, scikit-learn, joblib and seaborn); use the versions stated in the environment.yml file to reproduce the same environment used for this project.