官术网_书友最值得收藏!

Using the model for prediction

Let's say we have unseen data of TV ad spending and that we want to know their corresponding impact on the sales of the company. So, we need to use the learned model to do that for us. Let's suppose that we want to know how much sales will increase from $50000 of TV advertising.

Let's use our learned model coefficients to make such a calculation:

y = 7.032594 + 0.047537 x 50 

# manually calculating the increase in the sales based on $50k
7.032594 + 0.047537*50000

Output:

9,409.444

We can also use Statsmodels to make the prediction for us. First, we need to provide the TV ad value in a pandas DataFrame since the Statsmodels interface expects it:

# creating a Pandas DataFrame to match Statsmodels interface expectations
new_TVAdSpending = pd.DataFrame({'TV': [50000]})
new_TVAdSpending.head()

Output:

Now, we can go ahead and use the predict function to predict the sales value:

# use the model to make predictions on a new value
preds = lm.predict(new_TVAdSpending)

Output:

array([ 9.40942557])

Let's see how the learned least squares line looks. In order to draw the line, we need two points, with each point represented by this pair: (x, predict_value_of_x).

So, let's take the minimum and maximum values for the TV ad feature:

# create a DataFrame with the minimum and maximum values of TV
X_min_max = pd.DataFrame({'TV': [advertising_data.TV.min(), advertising_data.TV.max()]})
X_min_max.head()

Output:

Let's get the corresponding predictions for these two values:

# predictions for X min and max values
predictions = lm.predict(X_min_max)
predictions

Output:

array([  7.0658692,  21.12245377])

Now, let's plot the actual data and then fit it with the least squares line:

# plotting the acutal observed data
advertising_data.plot(kind='scatter', x='TV', y='sales')
#plotting the least squares line
plt.plot(new_TVAdSpending, preds, c='red', linewidth=2)

Output:

Figure 4: Plot of the actual data and the least squares line

Extensions of this example and further explanations will be explained in the next chapter.

主站蜘蛛池模板: 拜城县| 新巴尔虎右旗| 青海省| 榕江县| 保山市| 台北县| 上饶县| 云和县| 枣庄市| 盐城市| 交口县| 高阳县| 宁阳县| 肃南| 东丽区| 涞水县| 仪陇县| 辉县市| 龙胜| 宁津县| 八宿县| 上蔡县| 波密县| 伊春市| 桂东县| 清涧县| 仪陇县| 麟游县| 天等县| 泽普县| 宝鸡市| 竹北市| 涪陵区| 弋阳县| 巴塘县| 淮北市| 峨山| 梁平县| 通化县| 独山县| 神池县|