# What Is Two Feature Selection Strategies Chi Square An Information Gain

Decent Essays

2.3 Phase 3 (Feature selection) In our approach we utilized two feature selection strategies chi square an information gain. • Chi square: In our proposed system we utilized chi square as a scoring capacity with which we can discover if two terms are related to each other We at that point apply chi square capacity which gives the scoring capacity. Subsequent to applying chi square we learn whether the bigram or trigram happens as much of the time as every individual word. • Information gain: It causes us in comprehension if a word is educational or not. On the off chance that a word for the most part happens in positive survey and once in a while in negative audits it would main be able to that the word is vital. So we discover how …show more content…

The dashed line in the center demonstrates the choice edge of classifier. The regions set apart as FN and FP speaks to the inaccurately grouped items. Fig 2.2. Precision and Recall Precision and recall for the most part go up against each other and the spotted lines inside the bend speak to the choice limit for high recall or high precision separately. On the off chance that the choice limit is moved to one side, there will be a greater amount of FN objects and less FP objects, bringing about low recall and higher precision. Accuracy: Accuracy is the most widely recognized execution measure and it is a proportion of effectively anticipated perception to the aggregate perceptions. Accuracy is formalized: Accuracy (a)=(TP+TN)/(TP+FP+FN+TN) Precision (Completeness): Precision is the proportion of accurately anticipated positive perceptions to the aggregate anticipated positive perceptions. On the off chance that precision is high there will be low false positives. It is frequently restricting to recall as it is intuitive that lower recall give higher precision. Precision ( p)=TP/(TP+FP) Recall (Sensitivity): Recall is the proportion of accurately anticipated positive perceptions to the all perceptions in real class. Higher recalls relates to less false negatives as in condition Recall (r)=TP/(TP+FN) F-Measure: In factual examination, the F-measure (likewise F1 score or F-score) is a measure of the exactness. It considers both the precision