Parkinson’s disease (PD) is a central nervous system degenerative disease that affects the quality of life of millions of seniors around the world. Because of the variability of the condition, symptoms of Parkinson’s disease might progress differently from person to person. Patients with Parkinson’s disease may experience tremors, which occur mostly during rest. Tremors in the hands, limb rigidity, and gait and balance issues are all possibilities. Generally, two types of symptoms of PD can be distinguished: movement-related (i.e., motor) and unrelated to movement (non-motor). Patients with non-motor symptoms are actually more affected than those with motor symptoms. Depression, sleep behavior abnormalities, loss of smell, and cognitive impairment are examples of non-motor symptoms. It should be mentioned that early discovery of Parkinson’s disease allows for faster treatment and a significant reduction in symptoms. As a result, detecting PD at an early stage is critical for slowing its progression and, when available, may provide patients with the opportunity to receive disease-modifying therapy.
Visit The Below Link to Predict the Results.
Visit My GitHub link
STEP 1: Importing Libraries & Eda
- There are 195 rows and 24 columns
- remove the unwanted column
- There are no null values and no duplicates
- AND check the statistics of data and Plot the STATUS [Dependent Column] column.
- As we can see above there is imbalance in the data.
Do some Visualizations b/w dependent and independent variable. For Visualization You can visit my GitHub code.
STEP 2: Feature Engineering
For Feature selection we are using Variance Threshold
As we seen above code, we have removed some features by threshold value, we have finally 9 columns, next we can prepare our model.
STEP 3: Model Building (MACHINE LEARNING ALGORITHMS)
Before going through model building, we have to deal with imbalance of data, and we have to rename our columns which helps in deployment part.
Down below we have done, Imbalance and Normal Distribution
As above we have done balance of the data and normalization, x_train_new, y_train_new, is our new training model.
STEP 4: Machine Learning Algorithms
- SVM (Support Vector Machine)
We got 86% accuracy for SVM model.
2. Decision Tree Algorithm
we got 84% accuracy for decision tree classifier.
3. Random Forest Algorithm
we got 87% accuracy for random forest
4. XGB classifier
we got 89% accuracy for XGB algorithm.
We can increase our accuracy by performing Hyperparameter Tuning. We have finalized XGB as Final model.
Now Pickle The Model
STEP 5: Deployment
Above The code shows the deployment part.
STEP 6: Prediction
Open The Web APP
Give the Values ,
after giving values predict the values, down there is predict button.
Above is the predicted output for given values.
You can predict the PARKINSON’S DISEASE using below link
import the libraries
After running the code we got 87% accuracy.
Plotting the model loss and accuracy
Pickle the Model
As we seen above code, we have built model with machine learning algorithms and deep learning algorithms.