Machine learning algorithms and data visualisation for airfoil sound pressure level prediction

[amukh18]

Description

Your task is to predict the scaled sound pressure level(dB) from aerodynamic and acoustic tests of two and three-dimensional airfoil blade section conducted in an anechoic wind tunnel. This is a regression problem. An airfoil is a shaped surface such as an airplane wind, tail, or propeller blade that produces lift and drag when moved through the air. An anechoic wind tunnel is a type of wind tunnel that is designed to completely absorb sound waves.

The dataset can be found here.

The features convey the following information: 1. Frequency, in hertz. 2. Angle of attack(in degrees). 3. Chord length, in meters. 4. Free-stream velocity, in metres per second. 5. Suction side displacement thickness, in meters. 6. Scaled sound pressure level, in dB.

Details

• Technical Specifications: pandas, python, sci-kit, numpy
• Type of issue: Multiple
• Time Limit:
  • Plots must be implemented within 24 hours of being taken up.
  • Algorithms must be implemented within 2 days of being taken up.
  • Cross-validations/ensembles must be implemented within 3 days of being taken up.

The folder associated with the issue contains an IPython notebook containing a train-test split to get you started. It also contains the data set in CSV format.

Issue requirements / progress

All algorithms and ensembles must be scores using RMSE, Logloss and Accuracy metrics. Each pull request must only fulfill one of the tasks below.

Plots:

• [ ] Box plot (2 vs 4)
• [ ] Heatmap
• [ ] Linear regression model

Algorithms:

• [ ] Support vector machine
• [ ] K-nearest neighbors
• [ ] Random forest regressor
• [ ] XGBoost
• [ ] LightGBM
• [ ] Multi-layer Perceptron
• [ ] Decision Tree Regressor

Cross-validations/Ensembles:

• [ ] 10-fold cross-validation of XGBoost
• [ ] 10-fold cross-validation of LightGBM
• [ ] Average (non-weighted) the predictions of a 10-fold cross-validation of XGBoost and LightGBM (both must be implemented in the fold)

---

Resources

List of resources that might be required / helpful. Here are a few resources that may help you:

1. NumPy documentation: https://docs.scipy.org/doc/numpy-1.13.0/reference/index.html
2. Scikit-learn documentation: https://scikit-learn.org/stable/documentation.html
3. Pandas documentation: https://pandas.pydata.org/pandas-docs/stable/
4. Jupyter Notebook installation and tutorial : https://www.dataquest.io/blog/jupyter-notebook-tutorial/
5. XGBoost documentation: https://xgboost.readthedocs.io/en/latest/
6. LightGBM documentation: https://lightgbm.readthedocs.io/en/latest/
7. Scikit-learn documentation
   • Support vector machine: hhttps://scikit-learn.org/stable/modules/generated/sklearn.svm.SVR.html
   • K-nearest neighbors: https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.KNeighborsRegressor.html
   • Gaussian Naive Bayes: https://scikit-learn.org/stable/modules/generated/sklearn.naive_bayes.GaussianNB.html
   • Random forest regressor : https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html
   • Multi-layer perceptron regressor: https://scikit-learn.org/stable/modules/generated/sklearn.neural_network.MLPRegressor.html
   • Decision tree regressor: https://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeRegressor.html

8. Seaborn documentation a. Box plot: https://seaborn.pydata.org/generated/seaborn.boxplot.html b. Heat-map: http://seaborn.pydata.org/generated/seaborn.heatmap.html c. Linear regression model:https://seaborn.pydata.org/generated/seaborn.regplot.html

Directory Structure

The following convention must be adhered to when placing your solution files.

Plots:

• For Box-plot /machineLearning/airfoil/plots/bp/<solution_file>
• For Heatmap /machineLearning/airfoil/plots/hm/<solution_file>
• For Linear Regression Model /machineLearning/airfoil/plots/lrm/<solution_file>

Algorithms:

• For Support Vector Machine:
  /machineLearning/airfoil/algo/svm/<solution_file>
• For K-Nearest Neighbors: /machineLearning/airfoil/algo/knn/<solution_file>
• For Decision Trees Regressor: /machineLearning/airfoil/algo/dtm/<solution_file>
• For Random Forest Regressor: /machineLearning/airfoil/algo/rfm/<solution_file>
• For Multi-layer Perceptron: /machineLearning/airfoil/algo/mlp/<solution_file>
• For XGBoost: /machineLearning/airfoil/algo/xgb/<solution_file>
• For LightGBM: /machineLearning/airfoil/algo/lgbm/<solution_file>

Ensembles:

• For 10-fold XGBoost:
  /machineLearning/airfoil/ens/10-xgb/<solution_file>
• For 10-fold LightGBM: /machineLearning/airfoil/ens/10-lgbm/<solution_file>
• For average(non-weighted) of predictions of 10-fold XGBoost and 10-fold LightGBM /machineLearning/airfoil/ens/avg/<solution_file>

Note

Please claim the issue first by commenting here before starting to work on it. Feel free to contact @amukh18 or @CinnamonRolls1 with any issues at any time.