Applicazioni pratiche di machine learning/Previsioni sulle produzioni agricole: differenze tra le versioni
Contenuto cancellato Contenuto aggiunto
Nuova pagina: {{Applicazioni pratiche di machine learning}} ==Caricamento librerie== <syntaxhighlight lang="rsplus"> library(dplyr) library(caret) </syntaxhighlight> == Parte 1 : Dati == Il... |
|||
Riga 89:
</syntaxhighlight>
Call:
lm(formula = production ~ ., data = df)
Residuals:
Min 1Q Median 3Q Max
-983.8 -418.0 71.3 357.0 1103.3
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 2.714e+04 1.559e+03 17.415 2.16e-10 ***
x.1 -2.101e+03 1.428e+03 -1.471 0.16500
x.2 8.809e+02 4.152e+02 2.122 0.05365 .
x.3 -1.397e+02 5.110e+01 -2.735 0.01702 *
x.4 9.232e+00 2.787e+00 3.312 0.00561 **
x.5 -2.137e-01 5.551e-02 -3.851 0.00200 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 662.9 on 13 degrees of freedom
Multiple R-squared: 0.8649, Adjusted R-squared: 0.813
F-statistic: 16.65 on 5 and 13 DF, p-value: 2.999e-05
Dopo 2 anni secondo il modello regressivo la produzione di grano si dovrebbe azzerare :
Line 97 ⟶ 119:
predict.lm (model,newdata=df_pred, interval="confidence", level=0.80)
</syntaxhighlight>
fit lwr upr
1 12645.2107 10540.908 14749.513
2 -204.5818 -4776.246 4367.082
<syntaxhighlight lang="rsplus">
Line 113 ⟶ 140:
summary(model)
</syntaxhighlight>
Call:
lm(formula = production ~ ., data = df)
Residuals:
Min 1Q Median 3Q Max
-3022.94 -525.28 -8.86 991.11 2183.66
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.202e+04 5.175e+03 6.188 4.67e-05 ***
x.1 -1.380e+04 6.181e+03 -2.233 0.045393 *
x.2 6.588e+03 2.442e+03 2.698 0.019375 *
x.3 -1.364e+03 4.354e+02 -3.133 0.008643 **
x.4 1.372e+02 3.861e+01 3.553 0.003974 **
x.5 -6.563e+00 1.660e+00 -3.952 0.001919 **
x.6 1.196e-01 2.760e-02 4.333 0.000974 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 1491 on 12 degrees of freedom
Multiple R-squared: 0.9358, Adjusted R-squared: 0.9037
F-statistic: 29.14 on 6 and 12 DF, p-value: 1.754e-06
la produzione di miglio secondo il modello previsionale tenderà negli anni successivi a crescere nel seguente modo con un intervallo di confidenza all'80% :
Line 120 ⟶ 171:
predict.lm (model,newdata=df_pred, interval="confidence", level=0.80)
</syntaxhighlight>
fit lwr upr
1 77545.92 70527.07 84564.77
2 143944.02 124892.30 162995.75
3 260084.67 218097.79 302071.55
4 449274.05 367753.40 530794.69
5 741632.36 596404.25 886860.47
6 1175243.26 932368.21 1418118.32
7 1797389.41 1410612.72 2184166.10
8 2665874.07 2073687.17 3258060.98
9 3850428.88 2972713.36 4728144.41
10 5434207.62 4168434.96 6699980.29
11 7515366.14 5732325.90 9298406.38
<syntaxhighlight lang="rsplus">
| |||