{"id":88907,"date":"2021-05-21T12:17:57","date_gmt":"2021-05-21T16:17:57","guid":{"rendered":"https:\/\/ibkrcampus.com\/?p=88907"},"modified":"2024-05-20T15:57:09","modified_gmt":"2024-05-20T19:57:09","slug":"machine-learning-classification-algorithms-part-iii","status":"publish","type":"post","link":"https:\/\/www.interactivebrokers.com\/campus\/ibkr-quant-news\/machine-learning-classification-algorithms-part-iii\/","title":{"rendered":"Machine Learning Classification Algorithms \u2013 Part III"},"content":{"rendered":"\n

See Part I<\/a> and Part II<\/a> to get insight on Supervised Learning.<\/em><\/p>\n\n\n\n

Types of Classification<\/h2>\n\n\n\n

Based on the number and level of classes present in the dataset, there are three types of classification.<\/p>\n\n\n\n

Binary Classification<\/h3>\n\n\n\n

This type of classification has only two categories. Usually, they are boolean values – 1 or 0, True or False, High or Low. Some examples where such a classification could be used is in cancer detection or email spam detection where the labels would be positive or negative for cancer and spam or not spam for spam detection.<\/p>\n\n\n\n

Let us take an example. We are using a breast cancer detection dataset that can be downloaded from here<\/a>.<\/p>\n\n\n\n

\nimport pandas as pd
\nimport numpy as np
\nimport matplotlib.pyplot as plt
\nimport seaborn as sns
\ndata= pd.read_csv(“data.csv”)
\ndata.head()\n<\/p>\n\n\n\n

read_data_for_classification.py<\/a> hosted with \u2764 by GitHub<\/a><\/em><\/p>\n\n\n\n

<\/td>id<\/td>diagnosis<\/td>radius_mean<\/td>texture_mean<\/td>perimeter_mean<\/td>area_mean<\/td>smoothness_mean<\/td>compactness_mean<\/td>concavity_mean<\/td>concave points_mean<\/td>symmetry_mean<\/td>fractal_dimension_mean<\/td>radius_se<\/td>texture_se<\/td>perimeter_se<\/td>area_se<\/td>smoothness_se<\/td>compactness_se<\/td>concavity_se<\/td>concave points_se<\/td>symmetry_se<\/td>fractal_dimension_se<\/td>radius_worst<\/td>texture_worst<\/td>perimeter_worst<\/td>area_worst<\/td>smoothness_worst<\/td>compactness_worst<\/td>concavity_worst<\/td>concave points_worst<\/td>symmetry_worst<\/td>fractal_dimension_worst<\/td><\/tr><\/thead>
0<\/td>842302<\/td>M<\/td>17.99<\/td>10.38<\/td>122.80<\/td>1001.0<\/td>0.11840<\/td>0.27760<\/td>0.3001<\/td>0.14710<\/td>0.2419<\/td>0.07871<\/td>1.0950<\/td>0.9053<\/td>8.589<\/td>153.40<\/td>0.006399<\/td>0.04904<\/td>0.05373<\/td>0.01587<\/td>0.03003<\/td>0.006193<\/td>25.38<\/td>17.33<\/td>184.60<\/td>2019.0<\/td>0.1622<\/td>0.6656<\/td>0.7119<\/td>0.2654<\/td>0.4601<\/td>0.11890<\/td><\/tr>
1<\/td>842517<\/td>M<\/td>20.57<\/td>17.77<\/td>132.90<\/td>1326.0<\/td>0.08474<\/td>0.07864<\/td>0.0869<\/td>0.07017<\/td>0.1812<\/td>0.05667<\/td>0.5435<\/td>0.7339<\/td>3.398<\/td>74.08<\/td>0.005225<\/td>0.01308<\/td>0.01860<\/td>0.01340<\/td>0.01389<\/td>0.003532<\/td>24.99<\/td>23.41<\/td>158.80<\/td>1956.0<\/td>0.1238<\/td>0.1866<\/td>0.2416<\/td>0.1860<\/td>0.2750<\/td>0.08902<\/td><\/tr>
2<\/td>84300903<\/td>M<\/td>19.69<\/td>21.25<\/td>130.00<\/td>1203.0<\/td>0.10960<\/td>0.15990<\/td>0.1974<\/td>0.12790<\/td>0.2069<\/td>0.05999<\/td>0.7456<\/td>0.7869<\/td>4.585<\/td>94.03<\/td>0.006150<\/td>0.04006<\/td>0.03832<\/td>0.02058<\/td>0.02250<\/td>0.004571<\/td>23.57<\/td>25.53<\/td>152.50<\/td>1709.0<\/td>0.1444<\/td>0.4245<\/td>0.4504<\/td>0.2430<\/td>0.3613<\/td>0.08758<\/td><\/tr>
3<\/td>84348301<\/td>M<\/td>11.42<\/td>20.38<\/td>77.58<\/td>386.1<\/td>0.14250<\/td>0.28390<\/td>0.2414<\/td>0.10520<\/td>0.2597<\/td>0.09744<\/td>0.4956<\/td>1.1560<\/td>3.445<\/td>27.23<\/td>0.009110<\/td>0.07458<\/td>0.05661<\/td>0.01867<\/td>0.05963<\/td>0.009208<\/td>14.91<\/td>26.50<\/td>98.87<\/td>567.7<\/td>0.2098<\/td>0.8663<\/td>0.6869<\/td>0.2575<\/td>0.6638<\/td>0.17300<\/td><\/tr>
4<\/td>84358402<\/td>M<\/td>20.29<\/td>14.34<\/td>135.10<\/td>1297.0<\/td>0.10030<\/td>0.13280<\/td>0.1980<\/td>0.10430<\/td>0.1809<\/td>0.05883<\/td>0.7572<\/td>0.7813<\/td>5.438<\/td>94.44<\/td>0.011490<\/td>0.02461<\/td>0.05688<\/td>0.01885<\/td>0.01756<\/td>0.005115<\/td>22.54<\/td>16.67<\/td>152.20<\/td>1575.0<\/td>0.1374<\/td>0.2050<\/td>0.4000<\/td>0.1625<\/td>0.2364<\/td>0.07678<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

\nsns.scatterplot(x=”radius_mean”,y=”texture_mean”,hue=”diagnosis”,data=data)\n<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Fig. 2. Scatter Plot – Texture Mean vs. Radius Mean<\/p>\n\n\n\n

Here you can see the two \u2018classes\u2019 – \u2018M\u2019 stands for malignant and \u2018B\u2019 stands for benign. As you can see, the classes are well divided and are easily differentiable to the naked eye for these two features. However, this will not be true for all pairs of features.<\/p>\n\n\n\n

Models that can be used for such a classification are:<\/p>\n\n\n\n