Recomendaciones de cultivos
Irene Ayala
2025-07-18
Maximizar el rendimiento agrícola recomendando cultivos apropiados
La agricultura de precisión está de moda hoy en día. Ayuda a los agricultores a tomar decisiones informadas sobre su estrategia agrícola. Aquí les presento un conjunto de datos que permitirá a los usuarios crear un modelo predictivo para recomendar los cultivos más adecuados para una explotación agrícola específica, basándose en diversos parámetros.
Estructura de la informacion
Campos de datos:
N- Relación del contenido de nitrógeno en el suelo P- Relación del contenido de fósforo en el suelo K- Relación del contenido de potasio en el suelo Temperature- temperatura en grados Celsius Humidity- humedad relativa en % pH- valor de ph del suelo Rainfall- precipitación en mm
# Cargar librerías necesarias
library(knitr)
library(dplyr)
library(readr)
# Leer datos
datac <- read_csv("Crop_recommendation.csv")
# Ver estructura (opcional)
str(datac)## spc_tbl_ [2,200 × 8] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ N : num [1:2200] 90 85 60 74 78 69 69 94 89 68 ...
## $ P : num [1:2200] 42 58 55 35 42 37 55 53 54 58 ...
## $ K : num [1:2200] 43 41 44 40 42 42 38 40 38 38 ...
## $ temperature: num [1:2200] 20.9 21.8 23 26.5 20.1 ...
## $ humidity : num [1:2200] 82 80.3 82.3 80.2 81.6 ...
## $ ph : num [1:2200] 6.5 7.04 7.84 6.98 7.63 ...
## $ rainfall : num [1:2200] 203 227 264 243 263 ...
## $ label : chr [1:2200] "rice" "rice" "rice" "rice" ...
## - attr(*, "spec")=
## .. cols(
## .. N = col_double(),
## .. P = col_double(),
## .. K = col_double(),
## .. temperature = col_double(),
## .. humidity = col_double(),
## .. ph = col_double(),
## .. rainfall = col_double(),
## .. label = col_character()
## .. )
## - attr(*, "problems")=<externalptr># Calcular resumen estadístico solo para variables numéricas
summary_df <- datac %>%
select(where(is.numeric)) %>%
summary() %>%
as.data.frame.matrix() %>%
tibble::rownames_to_column("Estadística")
# Mostrar la tabla
kable(summary_df, caption = "Resumen estadístico de las variables numéricas")| Estadística | N | P | K | temperature | humidity | ph | rainfall |
|---|---|---|---|---|---|---|---|
| X | Min. : 0.00 | Min. : 5.00 | Min. : 5.00 | Min. : 8.826 | Min. :14.26 | Min. :3.505 | Min. : 20.21 |
| X.1 | 1st Qu.: 21.00 | 1st Qu.: 28.00 | 1st Qu.: 20.00 | 1st Qu.:22.769 | 1st Qu.:60.26 | 1st Qu.:5.972 | 1st Qu.: 64.55 |
| X.2 | Median : 37.00 | Median : 51.00 | Median : 32.00 | Median :25.599 | Median :80.47 | Median :6.425 | Median : 94.87 |
| X.3 | Mean : 50.55 | Mean : 53.36 | Mean : 48.15 | Mean :25.616 | Mean :71.48 | Mean :6.469 | Mean :103.46 |
| X.4 | 3rd Qu.: 84.25 | 3rd Qu.: 68.00 | 3rd Qu.: 49.00 | 3rd Qu.:28.562 | 3rd Qu.:89.95 | 3rd Qu.:6.924 | 3rd Qu.:124.27 |
| X.5 | Max. :140.00 | Max. :145.00 | Max. :205.00 | Max. :43.675 | Max. :99.98 | Max. :9.935 | Max. :298.56 |
freq_label <- datac %>%
count(label) %>%
arrange(desc(n))
kable(freq_label, col.names = c("Cultivo", "Frecuencia"), caption = "Frecuencia de cultivos en el conjunto de datos")| Cultivo | Frecuencia |
|---|---|
| apple | 100 |
| banana | 100 |
| blackgram | 100 |
| chickpea | 100 |
| coconut | 100 |
| coffee | 100 |
| cotton | 100 |
| grapes | 100 |
| jute | 100 |
| kidneybeans | 100 |
| lentil | 100 |
| maize | 100 |
| mango | 100 |
| mothbeans | 100 |
| mungbean | 100 |
| muskmelon | 100 |
| orange | 100 |
| papaya | 100 |
| pigeonpeas | 100 |
| pomegranate | 100 |
| rice | 100 |
| watermelon | 100 |
ggplot(freq_label, aes(x = reorder(label, -n), y = n)) +
geom_bar(stat = "identity", fill = "#2E86C1") +
labs(title = "Frecuencia de cultivos", x = "Cultivo", y = "Frecuencia") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
library(ggplot2)
library(gridExtra)
library(dplyr)
# Obtener nombres de variables numéricas
num_vars <- datac %>% select(where(is.numeric)) %>% names()
# Crear lista de boxplots
boxplots <- lapply(num_vars, function(var) {
ggplot(datac, aes_string(x = "label", y = var, fill = "label")) +
geom_boxplot(show.legend = FALSE, alpha = 0.7) +
labs(title = paste("Boxplot de", var, "por cultivo"), x = "Cultivo", y = var) +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
})
# Mostrar los boxplots en una cuadrícula de 2 columnas
do.call(grid.arrange, c(boxplots, ncol = 4))
##PCA (Análisis de Componentes Principales)
En la matriz de correlaciones la componente principal que explica mejor las varaibles de NPK es la CP1,las variables de temperatura, precipitacion y humedad no estan representadas.
library(stats)
data_scaled <- scale(datac[, 1:7])
#PCA
pca <- prcomp(data_scaled, center = TRUE, scale. = TRUE)
names(pca)## [1] "sdev" "rotation" "center" "scale" "x"head(pca$rotation)## PC1 PC2 PC3 PC4 PC5
## N -0.30219096 0.33410693 0.1120450 -0.54165059 0.50778466
## P 0.64378667 0.03435809 0.1099391 -0.04629318 -0.08233115
## K 0.62260719 0.28382920 0.1631733 -0.15486709 -0.03342452
## temperature -0.21242839 0.35948683 0.2482280 0.69082649 -0.15486542
## humidity -0.06848339 0.73791663 0.2135991 -0.06717140 -0.12887133
## ph -0.22694272 -0.22065738 0.5485203 -0.39570047 -0.65188053
## PC6 PC7
## N 0.48290443 0.008472888
## P 0.37684700 0.649104376
## K 0.02896707 -0.692268474
## temperature 0.50041798 -0.111281619
## humidity -0.54787098 0.289624027
## ph 0.12571195 -0.040027859dim(pca$rotation)## [1] 7 7pca$sdev^2## [1] 1.9312182 1.2939102 1.0765093 1.0228912 0.8059284 0.6765616 0.1929812summary(pca)## Importance of components:
## PC1 PC2 PC3 PC4 PC5 PC6 PC7
## Standard deviation 1.3897 1.1375 1.0375 1.0114 0.8977 0.82253 0.43930
## Proportion of Variance 0.2759 0.1848 0.1538 0.1461 0.1151 0.09665 0.02757
## Cumulative Proportion 0.2759 0.4607 0.6145 0.7607 0.8758 0.97243 1.00000#comprobar la importancia del componente 1
datac1 <- datac
compx1 <- pca$x
compx1 <- as.data.frame(compx1)
datac1$PC1 <- compx1$PC1
datac1$PC2 <- compx1$PC2
datac_cor <- subset(datac1, select = -c(label))
head(datac_cor)## # A tibble: 6 × 9
## N P K temperature humidity ph rainfall PC1 PC2
## <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 90 42 43 20.9 82.0 6.50 203. -0.583 0.844
## 2 85 58 41 21.8 80.3 7.04 227. -0.475 0.785
## 3 60 55 44 23.0 82.3 7.84 264. -0.634 0.694
## 4 74 35 40 26.5 80.2 6.98 243. -1.05 1.09
## 5 78 42 42 20.1 81.6 7.63 263. -0.873 0.659
## 6 69 37 42 23.1 83.4 7.07 251. -0.847 0.935cor(datac_cor)## N P K temperature humidity
## N 1.00000000 -0.23145958 -0.14051184 0.02650380 0.190688379
## P -0.23145958 1.00000000 0.73623222 -0.12754113 -0.118734116
## K -0.14051184 0.73623222 1.00000000 -0.16038713 0.190858861
## temperature 0.02650380 -0.12754113 -0.16038713 1.00000000 0.205319677
## humidity 0.19068838 -0.11873412 0.19085886 0.20531968 1.000000000
## ph 0.09668285 -0.13801889 -0.16950310 -0.01779502 -0.008482539
## rainfall 0.05902022 -0.06383905 -0.05346135 -0.03008378 0.094423053
## PC1 -0.41994957 0.89465925 0.86522649 -0.29520807 -0.095170191
## PC2 0.38004721 0.03908239 0.32285620 0.40891688 0.839381437
## ph rainfall PC1 PC2
## N 0.096682846 0.05902022 -4.199496e-01 3.800472e-01
## P -0.138018893 -0.06383905 8.946592e-01 3.908239e-02
## K -0.169503098 -0.05346135 8.652265e-01 3.228562e-01
## temperature -0.017795017 -0.03008378 -2.952081e-01 4.089169e-01
## humidity -0.008482539 0.09442305 -9.517019e-02 8.393814e-01
## ph 1.000000000 -0.10906948 -3.153784e-01 -2.509981e-01
## rainfall -0.109069484 1.00000000 -1.007959e-01 3.300552e-01
## PC1 -0.315378382 -0.10079595 1.000000e+00 5.627413e-16
## PC2 -0.250998148 0.33005522 5.627413e-16 1.000000e+00str(datac1)## spc_tbl_ [2,200 × 10] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ N : num [1:2200] 90 85 60 74 78 69 69 94 89 68 ...
## $ P : num [1:2200] 42 58 55 35 42 37 55 53 54 58 ...
## $ K : num [1:2200] 43 41 44 40 42 42 38 40 38 38 ...
## $ temperature: num [1:2200] 20.9 21.8 23 26.5 20.1 ...
## $ humidity : num [1:2200] 82 80.3 82.3 80.2 81.6 ...
## $ ph : num [1:2200] 6.5 7.04 7.84 6.98 7.63 ...
## $ rainfall : num [1:2200] 203 227 264 243 263 ...
## $ label : chr [1:2200] "rice" "rice" "rice" "rice" ...
## $ PC1 : num [1:2200] -0.583 -0.475 -0.634 -1.048 -0.873 ...
## $ PC2 : num [1:2200] 0.844 0.785 0.694 1.087 0.659 ...
## - attr(*, "spec")=
## .. cols(
## .. N = col_double(),
## .. P = col_double(),
## .. K = col_double(),
## .. temperature = col_double(),
## .. humidity = col_double(),
## .. ph = col_double(),
## .. rainfall = col_double(),
## .. label = col_character()
## .. )
## - attr(*, "problems")=<externalptr>#2 PCA en grafica y extraccion de la informacion
library(FactoMineR)
library(factoextra)
pcagr <- PCA(X = data_scaled, scale.unit = FALSE, ncp = 7, graph = FALSE)##Distribución de individuos por coseno cuadrado y de las variables
fviz_pca_ind(pcagr, col.ind = "cos2", gradient.cols= c("#00AFBB", "#E7B800" , "#FC4E07" ), repel = FALSE, graph = FALSE)
fviz_pca_var(pcagr, col.var = "cos2", geom.var = "arrow" , labelsize = 2 , repel = TRUE, graph = FALSE)
##Biplot
biplot(x = pca, scale = 0, cex = 0.7, col = c ("blue4", "brown3"))
##Modelo de entrenamiento y de test.
##Grafico de correlacion
index <- sample(1:nrow(datac), size = 0.7 * nrow(datac))
training_set <- datac[index, ]
testing_set <- datac[-index, ]
library(psych)
pairs.panels(training_set[ , -which(names(training_set) == "label")],
gap = 0,
bg = c("red", "yellow", "blue", "green", "orange")[as.factor(training_set$label)],
pch=21)
##PCA aplicado a las variables
pc <- prcomp(training_set[ , -which(names(training_set) == "label")],
center = TRUE,
scale. = TRUE)
summary(pc)## Importance of components:
## PC1 PC2 PC3 PC4 PC5 PC6 PC7
## Standard deviation 1.3996 1.1418 1.0369 1.0006 0.9003 0.81627 0.4291
## Proportion of Variance 0.2798 0.1862 0.1536 0.1430 0.1158 0.09518 0.0263
## Cumulative Proportion 0.2798 0.4661 0.6197 0.7627 0.8785 0.97370 1.0000##Biplot del PCA (componente 1 y 2)
library(ggbiplot)
bipdata <- ggbiplot(pc,
obs.scale = 1,
var.scale = 1,
groups = training_set$cultivo,
ellipse = TRUE,
circle = TRUE,
ellipse.prob = 0.68) +
scale_color_discrete(name = '') +
theme(legend.direction = 'horizontal',
legend.position = 'top')
print(bipdata)
###Anova para cada variable numerica
# Realizar ANOVA para cada variable numérica
variables_numericas <- c("N", "P", "K", "temperature", "humidity", "ph", "rainfall")
# Aplicar ANOVA y mostrar resultados
for (var in variables_numericas) {
formula <- as.formula(paste(var, "~ label"))
modelo <- aov(formula, data = datac)
cat("\n\n--- ANOVA para", var, "---\n")
print(summary(modelo))
}##
##
## --- ANOVA para N ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 2686561 127931 897.6 <2e-16 ***
## Residuals 2178 310433 143
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
##
## --- ANOVA para P ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 2267923 107996 1886 <2e-16 ***
## Residuals 2178 124740 57
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
##
## --- ANOVA para K ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 5619506 267596 27238 <2e-16 ***
## Residuals 2178 21397 10
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
##
## --- ANOVA para temperature ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 27984 1333 102.2 <2e-16 ***
## Residuals 2178 28402 13
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
##
## --- ANOVA para humidity ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 1054749 50226 3104 <2e-16 ***
## Residuals 2178 35246 16
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
##
## --- ANOVA para ph ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 484.5 23.070 60.34 <2e-16 ***
## Residuals 2178 832.7 0.382
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
##
## --- ANOVA para rainfall ---
## Df Sum Sq Mean Sq F value Pr(>F)
## label 21 5670650 270031 605.5 <2e-16 ***
## Residuals 2178 971264 446
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1modeloanova <- aov(N ~ label, data = datac)
tukey <- TukeyHSD(modeloanova)
plot(tukey, las = 1)
Modelo predictivo
library(devtools)
library(magrittr)
trg <- predict(pc, training_set) %>% as.data.frame()
trg$label <- training_set$label
tst <- predict(pc, testing_set) %>% as.data.frame()
tst$label <- testing_set$label
library(nnet)
trg$label <- as.factor(trg$label)
trg$label <- relevel(trg$label, ref = levels(trg$label)[1])
# Modelo multinomial con PC1 y PC2
modelo <- multinom(datac1$label ~ datac1$PC1 + datac1$PC2, datac1 = trg)## # weights: 88 (63 variable)
## initial value 6800.293397
## iter 10 value 4135.434734
## iter 20 value 3890.702543
## iter 30 value 3660.993176
## iter 40 value 3338.047015
## iter 50 value 2991.402434
## iter 60 value 2941.785238
## iter 70 value 2912.490009
## iter 80 value 2878.914009
## iter 90 value 2870.823220
## iter 100 value 2859.041959
## final value 2859.041959
## stopped after 100 iterationssummary(modelo)## Call:
## multinom(formula = datac1$label ~ datac1$PC1 + datac1$PC2, datac1 = trg)
##
## Coefficients:
## (Intercept) datac1$PC1 datac1$PC2
## banana 40.727238 -12.058426 -2.815998
## blackgram 49.713056 -25.190985 -25.523564
## chickpea 7.201289 -13.209071 -46.452630
## coconut 42.160176 -27.104980 -7.418631
## coffee 41.854712 -36.512966 -22.273522
## cotton 44.065073 -34.086528 -19.595812
## grapes 35.154403 -7.359129 -5.925505
## jute 47.004752 -25.678755 -11.833694
## kidneybeans 23.109845 -16.402953 -41.029263
## lentil 46.673791 -21.716308 -29.407807
## maize 49.601029 -26.613083 -24.621287
## mango 48.979361 -26.766745 -25.803705
## mothbeans 46.447966 -24.626284 -29.519902
## mungbean 50.545643 -25.733333 -22.322170
## muskmelon 37.526910 -33.388167 -8.938524
## orange 44.235649 -34.407060 -23.586795
## papaya 41.031339 -20.698092 -3.440360
## pigeonpeas 50.116822 -20.179157 -25.106984
## pomegranate 50.689055 -25.592215 -20.866910
## rice 42.659030 -24.018044 -5.937387
## watermelon 43.640741 -33.032157 -16.320874
##
## Std. Errors:
## (Intercept) datac1$PC1 datac1$PC2
## banana 6.291892 2.383462 3.4438080
## blackgram 6.344314 2.750023 3.6615933
## chickpea 8.599361 3.311182 4.5338686
## coconut 6.356990 2.730750 3.5551995
## coffee 6.377912 2.783081 3.6561107
## cotton 6.367177 2.769606 3.6462471
## grapes 5.190923 1.129208 0.8104685
## jute 6.347830 2.725398 3.5773158
## kidneybeans 8.256244 3.236289 4.4449524
## lentil 6.354444 2.760485 3.6782532
## maize 6.345069 2.750456 3.6594806
## mango 6.345681 2.751834 3.6601627
## mothbeans 6.352529 2.757397 3.6713668
## mungbean 6.343255 2.744109 3.6559485
## muskmelon 6.400201 2.787345 3.5994781
## orange 6.364231 2.766736 3.6557637
## papaya 6.348270 2.657828 3.5253223
## pigeonpeas 6.341493 2.747993 3.6763123
## pomegranate 6.342893 2.739012 3.6497814
## rice 6.348779 2.703728 3.5393187
## watermelon 6.368762 2.769795 3.6301405
##
## Residual Deviance: 5718.084
## AIC: 5844.084Matriz de confusion.
library(nnet)
pred_train <- predict(modelo, newdata = datac1)
# Matriz de confusión
tab_train <- table(pred_train, datac1$label)
print(tab_train)##
## pred_train apple banana blackgram chickpea coconut coffee cotton grapes jute
## apple 86 0 0 0 0 0 0 19 0
## banana 0 92 0 0 0 0 0 0 0
## blackgram 0 0 29 0 0 0 0 0 0
## chickpea 0 0 0 80 0 0 0 0 0
## coconut 0 0 0 0 30 0 1 0 14
## coffee 0 0 0 0 0 46 30 0 0
## cotton 0 0 0 0 0 17 27 0 1
## grapes 14 0 0 0 0 0 0 81 0
## jute 0 0 0 0 20 0 7 0 43
## kidneybeans 0 0 0 20 0 0 0 0 0
## lentil 0 0 3 0 0 0 0 0 0
## maize 0 0 14 0 0 0 0 0 0
## mango 0 0 21 0 0 1 0 0 0
## mothbeans 0 0 5 0 0 0 0 0 0
## mungbean 0 0 6 0 0 0 0 0 0
## muskmelon 0 0 0 0 20 2 5 0 3
## orange 0 0 3 0 0 25 4 0 0
## papaya 0 7 0 0 0 0 0 0 1
## pigeonpeas 0 0 15 0 0 0 0 0 0
## pomegranate 0 0 4 0 0 1 4 0 13
## rice 0 1 0 0 25 0 0 0 17
## watermelon 0 0 0 0 5 8 22 0 8
##
## pred_train kidneybeans lentil maize mango mothbeans mungbean muskmelon
## apple 0 0 0 0 0 0 0
## banana 0 0 0 0 0 0 0
## blackgram 0 2 21 25 5 8 0
## chickpea 25 0 0 0 0 0 0
## coconut 0 0 0 0 0 0 17
## coffee 0 0 0 0 0 0 0
## cotton 0 0 3 1 0 0 0
## grapes 0 0 0 0 0 0 0
## jute 0 0 0 0 0 0 2
## kidneybeans 73 0 0 0 1 0 0
## lentil 2 65 2 0 10 0 0
## maize 0 0 11 9 0 13 0
## mango 0 1 13 19 3 2 0
## mothbeans 0 18 8 15 50 0 0
## mungbean 0 0 17 7 1 30 0
## muskmelon 0 0 0 0 0 0 65
## orange 0 0 10 13 7 6 0
## papaya 0 0 0 0 0 0 0
## pigeonpeas 0 14 3 6 23 2 0
## pomegranate 0 0 12 5 0 39 0
## rice 0 0 0 0 0 0 0
## watermelon 0 0 0 0 0 0 16
##
## pred_train orange papaya pigeonpeas pomegranate rice watermelon
## apple 0 0 0 0 0 0
## banana 0 12 3 0 5 0
## blackgram 1 0 7 4 0 0
## chickpea 0 0 0 0 0 0
## coconut 1 1 0 0 14 0
## coffee 24 0 0 0 0 4
## cotton 7 0 0 3 0 21
## grapes 0 0 0 0 0 0
## jute 0 11 2 7 15 14
## kidneybeans 0 0 3 0 0 0
## lentil 0 0 21 0 0 0
## maize 4 0 2 8 0 0
## mango 12 0 1 1 0 0
## mothbeans 7 0 4 0 0 0
## mungbean 2 0 5 23 0 0
## muskmelon 4 0 0 0 8 14
## orange 30 0 0 7 0 0
## papaya 0 61 0 0 31 0
## pigeonpeas 0 0 46 2 0 0
## pomegranate 1 0 5 45 1 1
## rice 0 15 1 0 22 0
## watermelon 7 0 0 0 4 46# Error en entrenamiento
cat("Error entrenamiento: ", round((1 - sum(diag(tab_train)) / sum(tab_train)) * 100, 2), "%\n")## Error entrenamiento: 51.05 %pred_test <- predict(modelo, newdata = datac1)
tab_test <- table(Predicho = pred_test, Real = datac1$label)
print(tab_test)## Real
## Predicho apple banana blackgram chickpea coconut coffee cotton grapes jute
## apple 86 0 0 0 0 0 0 19 0
## banana 0 92 0 0 0 0 0 0 0
## blackgram 0 0 29 0 0 0 0 0 0
## chickpea 0 0 0 80 0 0 0 0 0
## coconut 0 0 0 0 30 0 1 0 14
## coffee 0 0 0 0 0 46 30 0 0
## cotton 0 0 0 0 0 17 27 0 1
## grapes 14 0 0 0 0 0 0 81 0
## jute 0 0 0 0 20 0 7 0 43
## kidneybeans 0 0 0 20 0 0 0 0 0
## lentil 0 0 3 0 0 0 0 0 0
## maize 0 0 14 0 0 0 0 0 0
## mango 0 0 21 0 0 1 0 0 0
## mothbeans 0 0 5 0 0 0 0 0 0
## mungbean 0 0 6 0 0 0 0 0 0
## muskmelon 0 0 0 0 20 2 5 0 3
## orange 0 0 3 0 0 25 4 0 0
## papaya 0 7 0 0 0 0 0 0 1
## pigeonpeas 0 0 15 0 0 0 0 0 0
## pomegranate 0 0 4 0 0 1 4 0 13
## rice 0 1 0 0 25 0 0 0 17
## watermelon 0 0 0 0 5 8 22 0 8
## Real
## Predicho kidneybeans lentil maize mango mothbeans mungbean muskmelon
## apple 0 0 0 0 0 0 0
## banana 0 0 0 0 0 0 0
## blackgram 0 2 21 25 5 8 0
## chickpea 25 0 0 0 0 0 0
## coconut 0 0 0 0 0 0 17
## coffee 0 0 0 0 0 0 0
## cotton 0 0 3 1 0 0 0
## grapes 0 0 0 0 0 0 0
## jute 0 0 0 0 0 0 2
## kidneybeans 73 0 0 0 1 0 0
## lentil 2 65 2 0 10 0 0
## maize 0 0 11 9 0 13 0
## mango 0 1 13 19 3 2 0
## mothbeans 0 18 8 15 50 0 0
## mungbean 0 0 17 7 1 30 0
## muskmelon 0 0 0 0 0 0 65
## orange 0 0 10 13 7 6 0
## papaya 0 0 0 0 0 0 0
## pigeonpeas 0 14 3 6 23 2 0
## pomegranate 0 0 12 5 0 39 0
## rice 0 0 0 0 0 0 0
## watermelon 0 0 0 0 0 0 16
## Real
## Predicho orange papaya pigeonpeas pomegranate rice watermelon
## apple 0 0 0 0 0 0
## banana 0 12 3 0 5 0
## blackgram 1 0 7 4 0 0
## chickpea 0 0 0 0 0 0
## coconut 1 1 0 0 14 0
## coffee 24 0 0 0 0 4
## cotton 7 0 0 3 0 21
## grapes 0 0 0 0 0 0
## jute 0 11 2 7 15 14
## kidneybeans 0 0 3 0 0 0
## lentil 0 0 21 0 0 0
## maize 4 0 2 8 0 0
## mango 12 0 1 1 0 0
## mothbeans 7 0 4 0 0 0
## mungbean 2 0 5 23 0 0
## muskmelon 4 0 0 0 8 14
## orange 30 0 0 7 0 0
## papaya 0 61 0 0 31 0
## pigeonpeas 0 0 46 2 0 0
## pomegranate 1 0 5 45 1 1
## rice 0 15 1 0 22 0
## watermelon 7 0 0 0 4 46cat("Error test: ", round((1 - sum(diag(tab_test)) / sum(tab_test)) * 100, 2), "%\n")## Error test: 51.05 %Randon Forest
library(readr)
library(dplyr)
library(caret)
library(randomForest)
library(nnet)
# Cargar y dividir los datos
set.seed(123)
datac <- read_csv("Crop_recommendation.csv")
# Separar en training (70%) y test (30%)
index <- createDataPartition(datac$label, p = 0.7, list = FALSE)
train1 <- datac[index, ]
test1 <- datac[-index, ]
train1$label <- as.factor(training_set$label)
test1$label <- as.factor(testing_set$label)
# Entrenar
modelo_rf <- randomForest(label ~ ., data = train1, ntree = 300)
# Predecir
pred_rf <- predict(modelo_rf, newdata = test1)
# Matriz de confusión
conf_rf <- confusionMatrix(pred_rf, test1$label)
print(conf_rf)## Confusion Matrix and Statistics
##
## Reference
## Prediction apple banana blackgram chickpea coconut coffee cotton grapes jute
## apple 0 0 6 2 0 1 0 1 0
## banana 0 2 1 1 6 1 4 1 0
## blackgram 1 1 1 0 1 0 6 1 8
## chickpea 0 2 1 0 1 3 0 3 2
## coconut 4 1 2 1 1 0 3 2 5
## coffee 3 0 2 5 1 0 0 0 1
## cotton 1 2 2 2 0 1 0 0 0
## grapes 2 0 2 1 0 3 0 6 5
## jute 2 1 1 5 0 2 1 0 0
## kidneybeans 1 5 2 0 0 1 4 0 1
## lentil 0 4 1 1 0 0 0 1 1
## maize 4 2 0 1 0 0 0 1 0
## mango 1 0 3 1 2 0 1 1 0
## mothbeans 0 0 5 0 2 0 1 2 0
## mungbean 1 3 0 1 3 0 0 1 0
## muskmelon 0 0 0 0 0 1 0 2 0
## orange 1 2 0 0 1 1 4 0 0
## papaya 1 0 0 2 0 5 1 3 0
## pigeonpeas 2 3 1 6 4 0 0 0 0
## pomegranate 0 0 1 4 0 8 0 0 0
## rice 0 0 3 0 1 0 1 1 1
## watermelon 1 0 5 0 0 2 1 0 2
## Reference
## Prediction kidneybeans lentil maize mango mothbeans mungbean muskmelon
## apple 2 1 1 3 1 5 2
## banana 1 0 0 2 0 1 0
## blackgram 0 0 1 0 0 2 0
## chickpea 0 2 1 1 1 1 0
## coconut 3 0 2 0 0 3 4
## coffee 2 3 0 0 2 0 7
## cotton 1 2 0 2 2 0 1
## grapes 2 0 3 0 0 1 1
## jute 4 1 2 3 0 3 1
## kidneybeans 0 1 3 2 1 0 1
## lentil 1 0 2 3 1 2 0
## maize 1 1 4 1 3 1 8
## mango 1 0 3 0 0 6 2
## mothbeans 3 9 2 5 5 3 1
## mungbean 4 4 0 6 0 2 0
## muskmelon 0 1 0 0 0 1 0
## orange 0 2 0 2 0 0 2
## papaya 1 3 1 0 6 1 3
## pigeonpeas 4 0 0 0 0 0 0
## pomegranate 0 0 3 1 1 0 2
## rice 0 4 1 1 5 3 2
## watermelon 2 1 1 0 1 1 2
## Reference
## Prediction orange papaya pigeonpeas pomegranate rice watermelon
## apple 1 0 2 1 0 0
## banana 2 3 0 0 0 1
## blackgram 1 0 0 1 5 0
## chickpea 1 0 0 1 0 4
## coconut 2 3 0 0 5 0
## coffee 2 0 0 1 2 6
## cotton 1 5 1 1 0 1
## grapes 0 0 2 0 0 2
## jute 6 2 0 0 0 0
## kidneybeans 1 0 2 1 2 0
## lentil 1 1 4 1 3 1
## maize 0 1 2 8 0 2
## mango 0 1 1 2 3 2
## mothbeans 0 0 0 5 3 1
## mungbean 1 3 3 1 0 0
## muskmelon 0 4 2 0 1 0
## orange 1 0 1 3 2 1
## papaya 1 0 1 1 2 1
## pigeonpeas 3 1 3 0 0 1
## pomegranate 3 0 2 1 0 0
## rice 4 1 2 3 1 0
## watermelon 1 2 0 0 0 1
##
## Overall Statistics
##
## Accuracy : 0.0424
## 95% CI : (0.0284, 0.0607)
## No Information Rate : 0.0591
## P-Value [Acc > NIR] : 0.9758
##
## Kappa : -0.0026
##
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: apple Class: banana Class: blackgram
## Sensitivity 0.00000 0.07143 0.025641
## Specificity 0.95433 0.96203 0.954911
## Pos Pred Value 0.00000 0.07692 0.034483
## Neg Pred Value 0.96038 0.95899 0.939778
## Prevalence 0.03788 0.04242 0.059091
## Detection Rate 0.00000 0.00303 0.001515
## Detection Prevalence 0.04394 0.03939 0.043939
## Balanced Accuracy 0.47717 0.51673 0.490276
## Class: chickpea Class: coconut Class: coffee Class: cotton
## Sensitivity 0.00000 0.043478 0.00000 0.00000
## Specificity 0.96172 0.937206 0.94136 0.96051
## Pos Pred Value 0.00000 0.024390 0.00000 0.00000
## Neg Pred Value 0.94811 0.964459 0.95345 0.95748
## Prevalence 0.05000 0.034848 0.04394 0.04091
## Detection Rate 0.00000 0.001515 0.00000 0.00000
## Detection Prevalence 0.03636 0.062121 0.05606 0.03788
## Balanced Accuracy 0.48086 0.490342 0.47068 0.48025
## Class: grapes Class: jute Class: kidneybeans Class: lentil
## Sensitivity 0.230769 0.00000 0.00000 0.00000
## Specificity 0.962145 0.94637 0.95541 0.95520
## Pos Pred Value 0.200000 0.00000 0.00000 0.00000
## Neg Pred Value 0.968254 0.95847 0.94937 0.94462
## Prevalence 0.039394 0.03939 0.04848 0.05303
## Detection Rate 0.009091 0.00000 0.00000 0.00000
## Detection Prevalence 0.045455 0.05152 0.04242 0.04242
## Balanced Accuracy 0.596457 0.47319 0.47771 0.47760
## Class: maize Class: mango Class: mothbeans Class: mungbean
## Sensitivity 0.133333 0.00000 0.172414 0.05556
## Specificity 0.942857 0.95223 0.933439 0.95032
## Pos Pred Value 0.100000 0.00000 0.106383 0.06061
## Neg Pred Value 0.958065 0.94921 0.960848 0.94577
## Prevalence 0.045455 0.04848 0.043939 0.05455
## Detection Rate 0.006061 0.00000 0.007576 0.00303
## Detection Prevalence 0.060606 0.04545 0.071212 0.05000
## Balanced Accuracy 0.538095 0.47611 0.552926 0.50294
## Class: muskmelon Class: orange Class: papaya
## Sensitivity 0.00000 0.031250 0.00000
## Specificity 0.98068 0.964968 0.94787
## Pos Pred Value 0.00000 0.043478 0.00000
## Neg Pred Value 0.93981 0.951334 0.95694
## Prevalence 0.05909 0.048485 0.04091
## Detection Rate 0.00000 0.001515 0.00000
## Detection Prevalence 0.01818 0.034848 0.05000
## Balanced Accuracy 0.49034 0.498109 0.47393
## Class: pigeonpeas Class: pomegranate Class: rice
## Sensitivity 0.107143 0.032258 0.034483
## Specificity 0.960443 0.960254 0.947702
## Pos Pred Value 0.107143 0.038462 0.029412
## Neg Pred Value 0.960443 0.952681 0.955272
## Prevalence 0.042424 0.046970 0.043939
## Detection Rate 0.004545 0.001515 0.001515
## Detection Prevalence 0.042424 0.039394 0.051515
## Balanced Accuracy 0.533793 0.496256 0.491092
## Class: watermelon
## Sensitivity 0.041667
## Specificity 0.965409
## Pos Pred Value 0.043478
## Neg Pred Value 0.963893
## Prevalence 0.036364
## Detection Rate 0.001515
## Detection Prevalence 0.034848
## Balanced Accuracy 0.503538varImpPlot(modelo_rf)
library(rpart)
library(rpart.plot)
# Crear modelo de árbol
modelo_arbol <- rpart(label ~ ., data = training_set, method = "class")
# Graficar
rpart.plot(modelo_arbol, type = 4, extra = 101, fallen.leaves = TRUE)
##Red Neuronal
library(nnet)
library(caret)
# Función de normalización
normalizar <- function(datac) {
df_norm <- datac
df_norm[, 1:7] <- scale(datac[, 1:7])
return(df_norm)
}
# Normalizar datos
train_n <- normalizar(training_set)
test_n <- normalizar(testing_set)
# Eliminar filas con NA
train_n <- na.omit(train_n)
test_n <- na.omit(test_n)
# Asegurar que label es factor
train_n$label <- as.factor(train_n$label)
test_n$label <- as.factor(test_n$label)
# Entrenar red neuronal
modelo_nnet <- nnet(label ~ ., data = train_n, size = 5, maxit = 500, trace = FALSE)
# Predecir sobre test
pred_nnet <- predict(modelo_nnet, newdata = test_n, type = "class")
# Asegurar que ambos vectores sean factores con mismos niveles
pred_nnet <- factor(pred_nnet, levels = levels(test_n$label))
test_n$label <- factor(test_n$label)
# Matriz de confusión
conf_nnet <- confusionMatrix(pred_nnet, test_n$label)
print(conf_nnet)## Confusion Matrix and Statistics
##
## Reference
## Prediction apple banana blackgram chickpea coconut coffee cotton grapes jute
## apple 25 0 0 0 0 0 0 0 0
## banana 0 25 0 0 0 0 0 0 0
## blackgram 0 0 39 0 0 0 0 0 0
## chickpea 0 0 0 33 0 0 0 0 0
## coconut 0 0 0 0 20 0 0 0 1
## coffee 0 0 0 0 0 29 0 0 0
## cotton 0 3 0 0 0 0 27 0 0
## grapes 0 0 0 0 0 0 0 26 0
## jute 0 0 0 0 0 0 0 0 15
## kidneybeans 0 0 0 0 0 0 0 0 0
## lentil 0 0 0 0 0 0 0 0 0
## maize 0 0 0 0 0 0 0 0 0
## mango 0 0 0 0 0 0 0 0 0
## mothbeans 0 0 0 0 0 0 0 0 0
## mungbean 0 0 0 0 0 0 0 0 0
## muskmelon 0 0 0 0 0 0 0 0 0
## orange 0 0 0 0 0 0 0 0 0
## papaya 0 0 0 0 3 0 0 0 0
## pigeonpeas 0 0 0 0 0 0 0 0 0
## pomegranate 0 0 0 0 0 0 0 0 0
## rice 0 0 0 0 0 0 0 0 10
## watermelon 0 0 0 0 0 0 0 0 0
## Reference
## Prediction kidneybeans lentil maize mango mothbeans mungbean muskmelon
## apple 0 0 0 0 0 0 1
## banana 0 0 0 0 0 0 0
## blackgram 0 4 0 0 1 1 0
## chickpea 0 0 0 0 0 0 0
## coconut 0 0 0 0 0 0 0
## coffee 0 0 3 0 0 0 0
## cotton 0 0 1 0 0 0 0
## grapes 0 0 0 0 0 0 0
## jute 0 0 0 0 0 0 0
## kidneybeans 26 0 0 0 0 0 0
## lentil 0 29 0 0 4 0 0
## maize 0 0 26 0 0 0 0
## mango 0 0 0 31 1 0 0
## mothbeans 0 2 0 0 23 0 0
## mungbean 0 0 0 0 0 35 0
## muskmelon 0 0 0 0 0 0 33
## orange 0 0 0 0 0 0 0
## papaya 0 0 0 0 0 0 1
## pigeonpeas 6 0 0 0 0 0 0
## pomegranate 0 0 0 0 0 0 0
## rice 0 0 0 0 0 0 0
## watermelon 0 0 0 1 0 0 4
## Reference
## Prediction orange papaya pigeonpeas pomegranate rice watermelon
## apple 0 0 0 0 0 0
## banana 0 0 0 0 0 0
## blackgram 0 0 0 0 0 0
## chickpea 0 0 0 0 0 0
## coconut 0 0 0 1 0 0
## coffee 0 0 1 0 0 0
## cotton 0 0 0 0 0 0
## grapes 0 0 0 0 0 0
## jute 0 0 0 0 0 0
## kidneybeans 0 0 0 0 0 0
## lentil 0 0 0 0 0 0
## maize 0 0 0 0 0 0
## mango 0 0 0 0 0 0
## mothbeans 0 0 0 0 0 0
## mungbean 0 0 0 0 0 0
## muskmelon 0 1 0 0 0 0
## orange 32 0 0 0 0 0
## papaya 0 26 0 1 0 0
## pigeonpeas 0 0 27 0 0 0
## pomegranate 0 0 0 28 0 0
## rice 0 0 0 0 29 0
## watermelon 0 0 0 1 0 24
##
## Overall Statistics
##
## Accuracy : 0.9212
## 95% CI : (0.898, 0.9406)
## No Information Rate : 0.0591
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9174
##
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: apple Class: banana Class: blackgram
## Sensitivity 1.00000 0.89286 1.00000
## Specificity 0.99843 1.00000 0.99034
## Pos Pred Value 0.96154 1.00000 0.86667
## Neg Pred Value 1.00000 0.99528 1.00000
## Prevalence 0.03788 0.04242 0.05909
## Detection Rate 0.03788 0.03788 0.05909
## Detection Prevalence 0.03939 0.03788 0.06818
## Balanced Accuracy 0.99921 0.94643 0.99517
## Class: chickpea Class: coconut Class: coffee Class: cotton
## Sensitivity 1.00 0.86957 1.00000 1.00000
## Specificity 1.00 0.99686 0.99366 0.99368
## Pos Pred Value 1.00 0.90909 0.87879 0.87097
## Neg Pred Value 1.00 0.99530 1.00000 1.00000
## Prevalence 0.05 0.03485 0.04394 0.04091
## Detection Rate 0.05 0.03030 0.04394 0.04091
## Detection Prevalence 0.05 0.03333 0.05000 0.04697
## Balanced Accuracy 1.00 0.93321 0.99683 0.99684
## Class: grapes Class: jute Class: kidneybeans Class: lentil
## Sensitivity 1.00000 0.57692 0.81250 0.82857
## Specificity 1.00000 1.00000 1.00000 0.99360
## Pos Pred Value 1.00000 1.00000 1.00000 0.87879
## Neg Pred Value 1.00000 0.98295 0.99054 0.99043
## Prevalence 0.03939 0.03939 0.04848 0.05303
## Detection Rate 0.03939 0.02273 0.03939 0.04394
## Detection Prevalence 0.03939 0.02273 0.03939 0.05000
## Balanced Accuracy 1.00000 0.78846 0.90625 0.91109
## Class: maize Class: mango Class: mothbeans Class: mungbean
## Sensitivity 0.86667 0.96875 0.79310 0.97222
## Specificity 1.00000 0.99841 0.99683 1.00000
## Pos Pred Value 1.00000 0.96875 0.92000 1.00000
## Neg Pred Value 0.99369 0.99841 0.99055 0.99840
## Prevalence 0.04545 0.04848 0.04394 0.05455
## Detection Rate 0.03939 0.04697 0.03485 0.05303
## Detection Prevalence 0.03939 0.04848 0.03788 0.05303
## Balanced Accuracy 0.93333 0.98358 0.89497 0.98611
## Class: muskmelon Class: orange Class: papaya
## Sensitivity 0.84615 1.00000 0.96296
## Specificity 0.99839 1.00000 0.99210
## Pos Pred Value 0.97059 1.00000 0.83871
## Neg Pred Value 0.99042 1.00000 0.99841
## Prevalence 0.05909 0.04848 0.04091
## Detection Rate 0.05000 0.04848 0.03939
## Detection Prevalence 0.05152 0.04848 0.04697
## Balanced Accuracy 0.92227 1.00000 0.97753
## Class: pigeonpeas Class: pomegranate Class: rice
## Sensitivity 0.96429 0.90323 1.00000
## Specificity 0.99051 1.00000 0.98415
## Pos Pred Value 0.81818 1.00000 0.74359
## Neg Pred Value 0.99841 0.99525 1.00000
## Prevalence 0.04242 0.04697 0.04394
## Detection Rate 0.04091 0.04242 0.04394
## Detection Prevalence 0.05000 0.04242 0.05909
## Balanced Accuracy 0.97740 0.95161 0.99208
## Class: watermelon
## Sensitivity 1.00000
## Specificity 0.99057
## Pos Pred Value 0.80000
## Neg Pred Value 1.00000
## Prevalence 0.03636
## Detection Rate 0.03636
## Detection Prevalence 0.04545
## Balanced Accuracy 0.99528###Comparacion de resultados
cat("Precision RF:", round(conf_rf$overall["Accuracy"]*100, 2), "%\n")## Precision RF: 4.24 %cat("Precision NNet:", round(conf_nnet$overall["Accuracy"]*100, 2), "%\n")## Precision NNet: 92.12 %###Resultado entre modelos: NNet (Red Neuronal) aprendio muy bien las relaciones entre las variables (N, P, K, temperatura, humedad, pH y lluvia) y el cultivo objetivo.
# Para red neuronal
accuracy_nnet <- conf_nnet$overall["Accuracy"]
kappa_nnet <- conf_nnet$overall["Kappa"]
precision_nnet <- conf_nnet$byClass[, "Precision"]
# Para RF
accuracy_rf <- conf_rf$overall["Accuracy"]
kappa_rf <- conf_rf$overall["Kappa"]
precision_rf <- conf_rf$byClass[, "Precision"]resultados <- data.frame(
Modelo = c("Red Neuronal", "Random Forest"),
Accuracy = c(accuracy_nnet, accuracy_rf),
Kappa = c(kappa_nnet, kappa_rf)
)
knitr::kable(resultados, caption = "Comparacion de modelos de clasificacion")| Modelo | Accuracy | Kappa |
|---|---|---|
| Red Neuronal | 0.9212121 | 0.9173876 |
| Random Forest | 0.0424242 | -0.0025501 |