第15章 回归的由来

本章包括下面各节：
15.1 平均数
15.2 向平均数方向的“回归”
15.3 线性回归

详细内容请阅读纸质书《Alink权威指南：基于Flink的机器学习实例入门（Python）》，这里为本章对应的示例代码。

from pyalink.alink import *
useLocalEnv(1)

from utils import *
import os
import pandas as pd

DATA_DIR = ROOT_DIR + "father_son" + os.sep

ORIGIN_FILE = "Pearson.txt";

source = CsvSourceBatchOp()\
    .setFilePath(DATA_DIR + ORIGIN_FILE)\
    .setSchemaStr("father double, son double")\
    .setFieldDelimiter("\t")\
    .setIgnoreFirstLine(True);

source.firstN(5).print();

import matplotlib.pyplot as plt

df_source = source.collectToDataframe()
plt.figure(figsize=(8, 8))
plt.scatter(df_source['father'], df_source['son'], color='blue', s=2)
plt.show()

source.lazyPrintStatistics();

BatchOperator.execute()

df_plus_one = source\
    .select("father, son, father+1 AS plus_one")\
    .collectToDataframe()
plt.figure(figsize=(8, 8))
plt.scatter(df_source['father'], df_source['son'], color='blue', s=2)
plt.plot(df_plus_one['father'], df_plus_one['plus_one'], color='grey', linewidth=2)
plt.show()

source.filter("father>=71.5 AND father<72.5").lazyPrintStatistics("father 72");

source.filter("father>=64.5 AND father<65.5").lazyPrintStatistics("father 65");

BatchOperator.execute()

linear_model = LinearRegTrainBatchOp()\
    .setFeatureCols(["father"])\
    .setLabelCol("son")\
    .linkFrom(source);

linear_model.lazyPrintTrainInfo();
linear_model.lazyPrintModelInfo();

linear_reg = LinearRegPredictBatchOp()\
    .setPredictionCol("linear_reg")\
    .linkFrom(linear_model, source);

linear_reg.lazyPrint(5);

BatchOperator.execute();

df_linear_reg = linear_reg.collectToDataframe()

plt.figure(figsize=(8, 8))
plt.scatter(df_source['father'], df_source['son'], color='blue', s=2)
plt.plot(df_plus_one['father'], df_plus_one['plus_one'], color='grey', linewidth=2)
plt.plot(df_linear_reg['father'], df_linear_reg['linear_reg'], color='red', linewidth=2)
plt.show()

```python

```