Group 2B: Web-based Machine Learning with Python,
consisting of:
Contents below are the same as: Basic-design-projects-update.pdf and README
The objective of this project is to design a web-based program (you can use this course for help: https://www.coursera.org/projects/machine-learning-streamlit-python) that, in addition to being user-friendly, includes the following tasks:
| No. | Step-by-step Guidethrough |
|---|---|
| 1. | Selecting input data (via CSV file) (Data.CSV) |
| 2. | Selecting independent and dependent data Independent data (all columns except the Flood column) Dependent data (Flood column) |
| 3. | Selecting the percentage ratio of training and testing data. The program should allow the selection of the following ratios for training and testing data: |
| o 90:10 (90% training data, 10% testing data) | |
| o 80:20 (80% training data, 20% testing data) | |
| o 70:30 (70% training data, 30% testing data) | |
| o 60:40 (60% training data, 40% testing data) | |
| 4. | Choosing a machine learning model for prediction on training and testing data: Machine learning models in regression mode (Random Forest or XGBoost) |
| 5. | Running the model on training and testing data and evaluating the results with the following metrics: |
RMSE |
|
MAE |
|
R2 |
|
| 6. | Displaying a histogram of errors between actual (Flood) and predicted data in step 5 |
| 7. | Displaying the determination of the importance of criteria using the machine learning model |
| 8. | Calling new data and predicting on this data: New data in CSV format is called initially. |
| This data has two columns x and y (longitude and latitude) and other columns similar to independent data in step 2. After predicting the model on this data (introducing independent columns) using the prediction output and values of x and y, display a density heat map here (you can use MapBox). |
| Extra Credits |
|---|
| o Steps 6 and 7 are considered extra credit. |
| o Choose a greater number of algorithms in step four. |
| o Display the area boundary in step 8 using https://mapshaper.org/ or Leaflet. |
- Import the Required Libraries:
from matplotlib.pyplot import figure
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from permetrics.regression import RegressionMetric
from sklearn.ensemble import RandomForestRegressor
import sklearn.metrics as metrics
- Load the Data:
ReadData = pd.read_csv('/content/drive/MyDrive/…../Data.csv')
- Split the Data into Training and Test Sets:
X = ReadData.drop([Flood], axis = 1)
y = ReadData [Flood]
4. Set the Data Ratio:
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, shuffle=True,
random_state=1)
- Create and Train the Model:
model = RandomForestRegressor(random_state = 1).fit(X_train, y_train)
- Predict Using the Model on Training and Test Data:
yhat_test = model.predict(X_test)
- Evaluate the Results for Training and Test Data:
r2 = metrics.r2_score(y_train, yhat_train)
mae = metrics.mean_absolute_error(y_train, yhat_train)
mse = metrics.mean_squared_error(y_train, yhat_train)
rmse = np.sqrt(mse)
r2 = metrics.r2_score(y_test, yhat_test)
mae = metrics.mean_absolute_error(y_test, yhat_test)
mse = metrics.mean_squared_error(y_test, yhat_test)
rmse = np.sqrt(mse)
- Load New Data:
NewData = pd.read_csv('/content/drive/MyDrive/…../NewData.csv')
- Predict Using the Model on New Data:
output_prediction = model.predict(Newdata)
- Calculate the Error Histogram:
plt.hist(error, bins=30)
plt.xlabel("Prediction Error")
plt.ylabel("Count")
plt.title("Error Histogram")
plt.show()
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent