Research Article
Breast Cancer Screening Using a Modified Inertial Projective Algorithms for Split Feasibility Problems
Table 2
Highest accuracy of ML algorithms after feature engineering.
| Machine learning model | Original dataset (%) | Highest accuracy (%) | Data cleaning or feature engineering |
| Logistic regression | 80.46 | 82.50 | Removing two least contributed features, RFE, or removing two least contributed features, correlation matrix | Linear discriminant analysis | 78.90 | 84.50 | Removing two least contributed features, RFE, or removing two least contributed features, correlation msatrix | K-nearest neighbors | 79.03 | 82.05 | Removing two least contributed features, RFE, or removing two least contributed features, correlation matrix | Classification and regression trees | 74.48 | 82.97 | Removing two least contributed features, RFE, or removing two least contributed features, correlation matrix | Gaussian Naive Bayes | 78.37 | 83.73 | Removing rows with missing values and outliers | Support vector machines | 80.34 | 83.73 | Removing two least contributed features, RFE, or removing two least contributed features, correlation matrix | Algorithm 1 () | 53.89 | 83.23 | Removing rows with missing values and outliers | Algorithm 1 () | 53.89 | 85.03 | Removing rows with missing values and outliers |
|
|