Category ➡️ Data Science
Subcategory ➡️ Data Scientist (Regression)
Difficulty ➡️ Easy
Objetivo del reto ➡️ Este reto ha sido creado específicamente para poner en práctica los conocimientos adquiridos en el curso: Machine Learning A-Z.
In the ever-evolving landscape of urban mobility, accurately forecasting the demand for bike-sharing services is crucial for planning and sustainability. BikeTech introduces the Bike Usage Prediction Challenge, inviting AI researchers to develop a predictive model capable of estimating the demand for bike-sharing services based on historical and environmental variables. This technology aims to optimize bike allocation, reduce operational costs, and enhance the user experience in urban bike-sharing programs.
Participants will be given access to the Bike Sharing Dataset, which includes historical bike usage data along with environmental and seasonal information:
dteday
, yr
, mnth
,hr
: Temporal data of bike usage.
weathersit
, temp
, atemp
, hum
, windspeed
: Weather conditions (temperature, humidity, wind speed).
season
:Seasonal information (spring, summer, fall, winter).
holiday
, weekday
, workingday
: Holiday and working day indicators.
cnt
: Historical usage patterns.
This dataset will be instrumental in training and evaluating your model's performance in predicting bike-sharing demand accurately.
Apply the appropriate data preprocessing techniques to handle the dataset's temporal aspects and environmental variables.
Choose and train a suitable predictive model. While there's no restriction on the type of model, techniques such as time series analysis, regression models, and machine learning algorithms like random forest or neural networks might prove effective.
The repository structure is provided and must be adhered to strictly:
|__README.md
|__requirements.txt
|
|__data
| |__train
| | |__train.csv
| |
| |__test
| |__test.csv
|
|__src
| |__data_processing.py
| |__model_training.py
| |__model_prediction.py
| |__utils.py
|
|__models
| |__model.pkl
|
|__predictions
|__example_predictions.json
|__predictions.json
The predictions
folder should contain the predictions.json
file with your model's predicted demand.
Task 1: Your task is to develop and train a predictive model that can accurately forecast the demand for bike-sharing services, contributing to BikeTech's mission of improving urban mobility solutions.
Submit a predictions.json file containing your model's demand forecasts. Ensure the file is correctly formatted, with the timestamp as the key and the predicted demand as the value. predictions.json:
{
"target": {
"2012-08-07 12:00": 23,
"2012-08-07 13:00": 52,
"2012-08-07 14:00": 312,
"2012-08-07 15:00": 11,
"2012-08-07 16:00": 125,
"2012-08-07 17:00": 642,
"2012-08-07 18:00": 76,
"2012-08-07 19:00": 53,
...
}
}
Performance will be assessed based on the accuracy of your demand forecasts. The Mean Absolute Percentage Error metrics will be used to evaluate the precision of your predictions.
⚠️ Please note:
All submissions might undergo a manual code review process to ensure that the work has been conducted honestly and adheres to the highest standards of academic integrity. Any form of dishonesty or misconduct will be addressed seriously, and may lead to disqualification from the challenge.
Ensure that all data manipulation and model training strictly utilize the libraries mentioned in requirements.txt.
Q1: What is the goal of the Bike Usage Prediction Challenge?
A1: To develop a model that can accurately predict the demand for bike-sharing services.
Q2: What type of data will I be working with?
A2: You will work with a dataset that includes historical bike usage data and environmental variables.
Q3: Are there any recommended models or algorithms?
A3: There's no strict requirement, but time series analysis, regression models, and machine learning algorithms like random forest or neural networks are suggested.
Q4: How will the model's performance be evaluated?
A4: Model performance will be evaluated using MAPE to measure the accuracy of demand forecasts.
Timeline
01
Start the challenge & clone the repository
02
Solve the challenge & submit your solution
Next action: