使用 sklearn.linearmodel 预测 x 值

Question

using sklearn to predict clearly gives wrong answer

this is my code:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn import linear_model

#read file
data=pd.read_csv(r"C:\Users\Administrator\OneDrive - Trøndelag fylkeskommune\Skrivebord\Scripts\POPULATION_NORWAY.csv")
data.head()

data=data [["YEAR","POPULATION"]]

print(data)

plt.scatter(data["YEAR"], data["POPULATION"] , color="blue")
plt.xlabel("YEAR")
plt.ylabel("POPULATION")
plt.show()

train=data[:(int((len(data))))]
test=data[(int((len(data)))):]

regr=linear_model.LinearRegression()

train_x=np.array(train[["YEAR"]])
train_y=np.array(train[["POPULATION"]])

regr.fit(train_x,train_y)

plt.scatter(train["YEAR"], train["POPULATION"], color="blue" )
plt.plot(train_x, regr.coef_*train_x+regr.intercept_, "-r")
plt.xlabel("Year")
plt.ylabel("Population")
plt.show()


def get_regression_predictions(input_features,intercept,slope):
    predicted_values=input_features*slope+intercept

    return predicted_values


future_year=float(input("what year is it?"))
future_population=float(input("what is the population?"))


estimated_population=get_regression_predictions(future_year,regr.intercept_[0],regr.coef_[0][0])
print ("estimated population :",estimated_population)


estimated_year=get_regression_predictions(future_population,regr.intercept_[0],regr.coef_[0][0])
print ("estimated year :",estimated_year)

these are the values:

YEAR  POPULATION
1900       2.217
1910       2.376
1920       2.616
1930       2.799
1940       2.963
1950       3.249
1960       3.567
1970       3.863
1980       4.078
1990       4.233
2000       4.478
2010       4.858
2020       5.367

this is population number in millions in Norway

this is the input and output:

what year is it?2050 

what is the population?6 

estimated population : 5.8595164835164795  

estimated year : -45.6942065934066 

as you can see, the estimated year is all wrong, i expect the estimated year to be someting closer to 2030

Answer 1

Well, you are approximating the population as function of time like this: Pop = year*slope + intercepet

To get the year in function of population, you need invert the expression:

year = (pop - intercept)/slope