/2-Creative-Deep-Learning-Models-Bike-Prediction

I am supper excited to share with you guys the results of new insights for the data preparation and building two models such as: Regression using SKLearn's Neural Network (NN) Train unisng Keras API with Tensorflow as Backend

Primary LanguageJupyter Notebook

2-Creative-Deep-Learning-Models-Bike-Prediction

I am supper excited to share with you guys the results of new insights for the data preparation and building two models such as: Regression using SKLearn's Neural Network (NN) Train unisng Keras API with Tensorflow as Backend

Kaggle Bike Sharing Demand Dataset

Bike sharing systems are a means of renting bicycles where the process of obtaining membership, rental, and bike return is automated via a network of kiosk locations throughout a city. Using these systems, people are able rent a bike from a one location and return it to a different place on an as-needed basis. Currently, there are over 500 bike-sharing programs around the world.

The data generated by these systems makes them attractive for researchers because the duration of travel, departure location, arrival location, and time elapsed is explicitly recorded. Bike sharing systems therefore function as a sensor network, which can be used for studying mobility in a city. In this competition, participants are asked to combine historical usage patterns with weather data in order to forecast bike rental demand in the Capital Bikeshare program in Washington, D.C.

bikes

Modified 'count' to log1p(count) for training

Log can be used when target represents a count (that is non-negative values)

Model now predicts as log1p(count). We need to convert it back to actual count using expm1(predicted_target)

Input Features: ['season', 'holiday', 'workingday', 'weather', 'temp', 'atemp', 'humidity', 'windspeed', 'year', 'month', 'day', 'dayofweek','hour'] Target Feature: [log1p('count')]

Objective:

We are provided hourly rental data spanning two years. For this competition, the training set is comprised of the first 19 days of each month, while the test set is the 20th to the end of the month. You must predict the total count of bikes rented during each hour covered by the test set, using only information available prior to the rental period (Ref: Kaggle.com)

In this Notebook, we will go through some steps and different insights such as:

Contains

Regression using SKLearn's Neural Network (NN)

Data Underestanding

Data Visualization

Data Preparation:

One Hot Encode all the Categorical Features

Standardize or Normalize all the Numeric Features

Train using SKLearn's MLPRegressor (Multi-Layer Perceptron)/Regression using SKLearn's Neural Network (NN)

Prediction

Evaluating the Results

Submission # New Insight for Modeling # Train unisng Keras API with Tensorflow as Backend

Data Underestanding

Data Visualization

Data Preparation:

One Hot Encode all the Categorical Features

Standardize or Normalize all the Numeric Features

Train unisng Keras API with Tensorflow as Backend

Prediction

Evaluating the Results

Submission

The Accuracy of our model could be shown as below:

accuracy

Second Part

NEW Insight

Regression using TensorFlow

Build the Neural Network using Keras - Easy and Portable across different implementations https://keras.io/

Objective:

Train a bike rental prediction model

NN requires one hot encoding of categorical data

NN also requires features to be on similar scale

Perform one-hot encoding of all categorical features: ['season', 'holiday', 'workingday', 'weather', 'year', 'month', 'day', 'dayofweek', 'hour'] Verify model performance

The accuracy of second model could be shown as below:

accuracy

for more information, check out my notebook on my Kaggle Profile here:

https://www.kaggle.com/homayoonkhadivi/2-creative-deep-learning-models-bike-prediction