forked from runawayhorse001/learning-apache-spark
-
Notifications
You must be signed in to change notification settings - Fork 0
/
pyspark.ml.feature-module.Rmd
28 lines (20 loc) · 1.14 KB
/
pyspark.ml.feature-module.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
---
title: "pyspark.ml.feature module"
author: "Wenqiang Feng & Ming Chen"
date: "2/15/2017"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```
## Introduction
This module provides a set of functions, methods and classes which act on **features**. A feature is like a column from the data frame or table. You can see that most of functions or classes take parameters like `inputCol`, `featuresCol`, `outputCol`, `labelCol`. These parameters specify the names of column (features) that you want to work on.
## class pairs and `fit/transform` functions
I found that there are a lot of class pairs in this module. For example:
* `ChiSqSelector` and `ChiSqSelectorModel`
* `CountVectorizer` and `CountVectorizerModel`
* `IDF` and `IDFModel`
* a lot of other pairs ...
The first class in a pair have functions to build model (instructions about how you want to transform your data). The second class in a pair do the actual data transformation.
* The `fit` function is from the first class and fit the built model to your data.
* The `transform` function is from the second class and does the actual data transformation.