- Neural Networks with Keras Cookbook
- V Kishore Ayyadevara
- 184字
- 2021-07-02 12:46:31
How it works...
In this recipe, we have learned about the following concepts:
- Imputing missing values: We have learned that one of the ways to impute the missing values of a variable is by replacing the missing values with the median of the corresponding variable. Other ways to deal with the missing values is by replacing them with the mean value, and also by replacing the missing value with the mean of the variable's value in the rows that are most similar to the row that contains a missing value (this technique is called identifying the K-Nearest Neighbours).
- Capping the outlier values: We have also learned that one way to cap the outliers is by replacing values that are above the 95th percentile value with the 95th percentile value. The reason we performed this exercise is to ensure that the input variable does not have all the values clustered around a small value (when the variable is scaled by the maximum value, which is an outlier).
- Scaling dataset: Finally, we scaled the dataset so that it can then be passed to a neural network.
推薦閱讀
- Getting Started with Gulp(Second Edition)
- 程序員面試筆試寶典(第3版)
- C語(yǔ)言程序設(shè)計(jì)(第3版)
- PHP基礎(chǔ)案例教程
- NLTK基礎(chǔ)教程:用NLTK和Python庫(kù)構(gòu)建機(jī)器學(xué)習(xí)應(yīng)用
- 數(shù)據(jù)結(jié)構(gòu)(Java語(yǔ)言描述)
- Learning Bayesian Models with R
- Neo4j Essentials
- QTP自動(dòng)化測(cè)試進(jìn)階
- Python忍者秘籍
- Flutter跨平臺(tái)開(kāi)發(fā)入門與實(shí)戰(zhàn)
- Learning Laravel's Eloquent
- Raspberry Pi Home Automation with Arduino(Second Edition)
- Python+Tableau數(shù)據(jù)可視化之美
- 區(qū)塊鏈技術(shù)進(jìn)階與實(shí)戰(zhàn)(第2版)