Importar base de datos

#file.choose()

stores <- read.csv("/Users/vanessaelizondo/Desktop/Tec/Semestre 7/Walmart/stores.csv")
features <- read.csv("/Users/vanessaelizondo/Desktop/Tec/Semestre 7/Walmart/features.csv")
train <- read.csv("/Users/vanessaelizondo/Desktop/Tec/Semestre 7/Walmart/train.csv")
test <- read.csv("/Users/vanessaelizondo/Desktop/Tec/Semestre 7/Walmart/test.csv")

Instalar paquetes

#install.packages("dplyr")
library(dplyr)
## 
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

Entender las bases de datos

summary(stores)
##      Store        Type                Size       
##  Min.   : 1   Length:45          Min.   : 34875  
##  1st Qu.:12   Class :character   1st Qu.: 70713  
##  Median :23   Mode  :character   Median :126512  
##  Mean   :23                      Mean   :130288  
##  3rd Qu.:34                      3rd Qu.:202307  
##  Max.   :45                      Max.   :219622
count(stores, Type, sort=TRUE)
##   Type  n
## 1    A 22
## 2    B 17
## 3    C  6
str(stores)
## 'data.frame':    45 obs. of  3 variables:
##  $ Store: int  1 2 3 4 5 6 7 8 9 10 ...
##  $ Type : chr  "A" "A" "B" "A" ...
##  $ Size : int  151315 202307 37392 205863 34875 202505 70713 155078 125833 126512 ...
summary(features)
##      Store        Date            Temperature       Fuel_Price   
##  Min.   : 1   Length:8190        Min.   : -7.29   Min.   :2.472  
##  1st Qu.:12   Class :character   1st Qu.: 45.90   1st Qu.:3.041  
##  Median :23   Mode  :character   Median : 60.71   Median :3.513  
##  Mean   :23                      Mean   : 59.36   Mean   :3.406  
##  3rd Qu.:34                      3rd Qu.: 73.88   3rd Qu.:3.743  
##  Max.   :45                      Max.   :101.95   Max.   :4.468  
##                                                                  
##    MarkDown1        MarkDown2           MarkDown3           MarkDown4       
##  Min.   : -2781   Min.   :  -265.76   Min.   :  -179.26   Min.   :    0.22  
##  1st Qu.:  1578   1st Qu.:    68.88   1st Qu.:     6.60   1st Qu.:  304.69  
##  Median :  4744   Median :   364.57   Median :    36.26   Median : 1176.42  
##  Mean   :  7032   Mean   :  3384.18   Mean   :  1760.10   Mean   : 3292.94  
##  3rd Qu.:  8923   3rd Qu.:  2153.35   3rd Qu.:   163.15   3rd Qu.: 3310.01  
##  Max.   :103185   Max.   :104519.54   Max.   :149483.31   Max.   :67474.85  
##  NA's   :4158     NA's   :5269        NA's   :4577        NA's   :4726      
##    MarkDown5             CPI         Unemployment    IsHoliday      
##  Min.   :  -185.2   Min.   :126.1   Min.   : 3.684   Mode :logical  
##  1st Qu.:  1440.8   1st Qu.:132.4   1st Qu.: 6.634   FALSE:7605     
##  Median :  2727.1   Median :182.8   Median : 7.806   TRUE :585      
##  Mean   :  4132.2   Mean   :172.5   Mean   : 7.827                  
##  3rd Qu.:  4832.6   3rd Qu.:213.9   3rd Qu.: 8.567                  
##  Max.   :771448.1   Max.   :229.0   Max.   :14.313                  
##  NA's   :4140       NA's   :585     NA's   :585
str(features)
## 'data.frame':    8190 obs. of  12 variables:
##  $ Store       : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Date        : chr  "05/02/2010" "12/02/2010" "19/02/2010" "26/02/2010" ...
##  $ Temperature : num  42.3 38.5 39.9 46.6 46.5 ...
##  $ Fuel_Price  : num  2.57 2.55 2.51 2.56 2.62 ...
##  $ MarkDown1   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown2   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown3   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown4   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown5   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ CPI         : num  211 211 211 211 211 ...
##  $ Unemployment: num  8.11 8.11 8.11 8.11 8.11 ...
##  $ IsHoliday   : logi  FALSE TRUE FALSE FALSE FALSE FALSE ...

Observaciones

  1. En”features” y “test” tienen la fecha como carácter.
  2. En “features” , hay NA´s en más de la mitad de los registros de Markdown (del 1 al 15)
  3. En “features” , hay 585 NA´s en CPI, Unemployement y hay 585 registros de Holiday = TRUE ¿Tiene relación? R: Sin relación
  4. En “train” , hay ventas semanales negativas.

Herramienta “El Generador de Valor de Datos”

Paso 1. Definir el area del negocio que buscamos impactar o mejorar y su KPI.

#El deparatamento de mercadotecnia de EUA (con muestra de 45 tiendas) en el indicador de ventas semanales.

Paso 2. Seleccionar plantilla (-s) para crear valor a partir de los datos de los clientes.

Visión / Segmentación / Personalización / Contextualización

Paso 3. Generar Ideas o conceptos específicos

#Elaborar un modelo de ventas predictivas

Paso 4. Reunir los datos requeridos

#Elaborar una base de datos con la variable dependiente (Ventas semanales) y las variables independientes.

Paso 5. Plan de ejecución

#Mercadotecnia elabora plan para desplegar eel modelo preditivo en fases:

#Fase 1. Piloto (San Antonio, Tx)
#Fase 2. Texas
#Fase 3. EUA

#Sistemas asegurara la captura del Markdown en las bases de datos.

Creacion de la base de datos consolidada

Agregar “stores” a “train”

bd<- merge(train,stores, by= "Store")

Agregar “features” a “bd”

bd1<- bd
bd1<- merge(bd1,features)

Eliminar columnas

bd2<- bd1
str(bd2)
## 'data.frame':    421570 obs. of  16 variables:
##  $ Store       : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Date        : chr  "01/04/2011" "01/04/2011" "01/04/2011" "01/04/2011" ...
##  $ IsHoliday   : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
##  $ Dept        : int  49 26 81 34 59 30 7 85 8 28 ...
##  $ Weekly_Sales: num  13168 5947 28545 9950 317 ...
##  $ Type        : chr  "A" "A" "A" "A" ...
##  $ Size        : int  151315 151315 151315 151315 151315 151315 151315 151315 151315 151315 ...
##  $ Temperature : num  59.2 59.2 59.2 59.2 59.2 ...
##  $ Fuel_Price  : num  3.52 3.52 3.52 3.52 3.52 ...
##  $ MarkDown1   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown2   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown3   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown4   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ MarkDown5   : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ CPI         : num  215 215 215 215 215 ...
##  $ Unemployment: num  7.68 7.68 7.68 7.68 7.68 ...
bd2<-subset(bd2, select=-c(MarkDown1,MarkDown2,MarkDown3,MarkDown4,MarkDown5))
summary(bd2)
##      Store          Date           IsHoliday            Dept      
##  Min.   : 1.0   Length:421570      Mode :logical   Min.   : 1.00  
##  1st Qu.:11.0   Class :character   FALSE:391909    1st Qu.:18.00  
##  Median :22.0   Mode  :character   TRUE :29661     Median :37.00  
##  Mean   :22.2                                      Mean   :44.26  
##  3rd Qu.:33.0                                      3rd Qu.:74.00  
##  Max.   :45.0                                      Max.   :99.00  
##   Weekly_Sales        Type                Size         Temperature    
##  Min.   : -4989   Length:421570      Min.   : 34875   Min.   : -2.06  
##  1st Qu.:  2080   Class :character   1st Qu.: 93638   1st Qu.: 46.68  
##  Median :  7612   Mode  :character   Median :140167   Median : 62.09  
##  Mean   : 15981                      Mean   :136728   Mean   : 60.09  
##  3rd Qu.: 20206                      3rd Qu.:202505   3rd Qu.: 74.28  
##  Max.   :693099                      Max.   :219622   Max.   :100.14  
##    Fuel_Price         CPI         Unemployment   
##  Min.   :2.472   Min.   :126.1   Min.   : 3.879  
##  1st Qu.:2.933   1st Qu.:132.0   1st Qu.: 6.891  
##  Median :3.452   Median :182.3   Median : 7.866  
##  Mean   :3.361   Mean   :171.2   Mean   : 7.960  
##  3rd Qu.:3.738   3rd Qu.:212.4   3rd Qu.: 8.572  
##  Max.   :4.468   Max.   :227.2   Max.   :14.313

Cambiar formato de fecha

bd2$Date <- as.Date(bd2$Date, format= "%d/%m/%Y")
str(bd2)
## 'data.frame':    421570 obs. of  11 variables:
##  $ Store       : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Date        : Date, format: "2011-04-01" "2011-04-01" ...
##  $ IsHoliday   : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
##  $ Dept        : int  49 26 81 34 59 30 7 85 8 28 ...
##  $ Weekly_Sales: num  13168 5947 28545 9950 317 ...
##  $ Type        : chr  "A" "A" "A" "A" ...
##  $ Size        : int  151315 151315 151315 151315 151315 151315 151315 151315 151315 151315 ...
##  $ Temperature : num  59.2 59.2 59.2 59.2 59.2 ...
##  $ Fuel_Price  : num  3.52 3.52 3.52 3.52 3.52 ...
##  $ CPI         : num  215 215 215 215 215 ...
##  $ Unemployment: num  7.68 7.68 7.68 7.68 7.68 ...
#install.packages("wordspace")
library(wordspace)
## Loading required package: Matrix
signcount(bd2$Weekly_Sales)
##    pos   zero    neg 
## 420212     73   1285

Eliminar ventas menores que 0

bd3 <- bd2
bd3 <- bd3[bd3$Weekly_Sales > 0,]
summary (bd3)
##      Store           Date            IsHoliday            Dept      
##  Min.   : 1.0   Min.   :2010-02-05   Mode :logical   Min.   : 1.00  
##  1st Qu.:11.0   1st Qu.:2010-10-08   FALSE:390652    1st Qu.:18.00  
##  Median :22.0   Median :2011-06-17   TRUE :29560     Median :37.00  
##  Mean   :22.2   Mean   :2011-06-18                   Mean   :44.24  
##  3rd Qu.:33.0   3rd Qu.:2012-02-24                   3rd Qu.:74.00  
##  Max.   :45.0   Max.   :2012-10-26                   Max.   :99.00  
##   Weekly_Sales        Type                Size         Temperature    
##  Min.   :     0   Length:420212      Min.   : 34875   Min.   : -2.06  
##  1st Qu.:  2120   Class :character   1st Qu.: 93638   1st Qu.: 46.68  
##  Median :  7662   Mode  :character   Median :140167   Median : 62.09  
##  Mean   : 16033                      Mean   :136750   Mean   : 60.09  
##  3rd Qu.: 20271                      3rd Qu.:202505   3rd Qu.: 74.28  
##  Max.   :693099                      Max.   :219622   Max.   :100.14  
##    Fuel_Price         CPI         Unemployment   
##  Min.   :2.472   Min.   :126.1   Min.   : 3.879  
##  1st Qu.:2.933   1st Qu.:132.0   1st Qu.: 6.891  
##  Median :3.452   Median :182.4   Median : 7.866  
##  Mean   :3.361   Mean   :171.2   Mean   : 7.960  
##  3rd Qu.:3.738   3rd Qu.:212.4   3rd Qu.: 8.567  
##  Max.   :4.468   Max.   :227.2   Max.   :14.313

Agregar numero de la semana

bd4 <- bd3
bd4$week_number <- strftime(bd4$Date, format = "%V")
str(bd4)
## 'data.frame':    420212 obs. of  12 variables:
##  $ Store       : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Date        : Date, format: "2011-04-01" "2011-04-01" ...
##  $ IsHoliday   : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
##  $ Dept        : int  49 26 81 34 59 30 7 85 8 28 ...
##  $ Weekly_Sales: num  13168 5947 28545 9950 317 ...
##  $ Type        : chr  "A" "A" "A" "A" ...
##  $ Size        : int  151315 151315 151315 151315 151315 151315 151315 151315 151315 151315 ...
##  $ Temperature : num  59.2 59.2 59.2 59.2 59.2 ...
##  $ Fuel_Price  : num  3.52 3.52 3.52 3.52 3.52 ...
##  $ CPI         : num  215 215 215 215 215 ...
##  $ Unemployment: num  7.68 7.68 7.68 7.68 7.68 ...
##  $ week_number : chr  "13" "13" "13" "13" ...
bd4$week_number <- as.integer(bd4$week_number)
str(bd4)
## 'data.frame':    420212 obs. of  12 variables:
##  $ Store       : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Date        : Date, format: "2011-04-01" "2011-04-01" ...
##  $ IsHoliday   : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
##  $ Dept        : int  49 26 81 34 59 30 7 85 8 28 ...
##  $ Weekly_Sales: num  13168 5947 28545 9950 317 ...
##  $ Type        : chr  "A" "A" "A" "A" ...
##  $ Size        : int  151315 151315 151315 151315 151315 151315 151315 151315 151315 151315 ...
##  $ Temperature : num  59.2 59.2 59.2 59.2 59.2 ...
##  $ Fuel_Price  : num  3.52 3.52 3.52 3.52 3.52 ...
##  $ CPI         : num  215 215 215 215 215 ...
##  $ Unemployment: num  7.68 7.68 7.68 7.68 7.68 ...
##  $ week_number : int  13 13 13 13 13 13 13 13 13 13 ...
summary(bd4)
##      Store           Date            IsHoliday            Dept      
##  Min.   : 1.0   Min.   :2010-02-05   Mode :logical   Min.   : 1.00  
##  1st Qu.:11.0   1st Qu.:2010-10-08   FALSE:390652    1st Qu.:18.00  
##  Median :22.0   Median :2011-06-17   TRUE :29560     Median :37.00  
##  Mean   :22.2   Mean   :2011-06-18                   Mean   :44.24  
##  3rd Qu.:33.0   3rd Qu.:2012-02-24                   3rd Qu.:74.00  
##  Max.   :45.0   Max.   :2012-10-26                   Max.   :99.00  
##   Weekly_Sales        Type                Size         Temperature    
##  Min.   :     0   Length:420212      Min.   : 34875   Min.   : -2.06  
##  1st Qu.:  2120   Class :character   1st Qu.: 93638   1st Qu.: 46.68  
##  Median :  7662   Mode  :character   Median :140167   Median : 62.09  
##  Mean   : 16033                      Mean   :136750   Mean   : 60.09  
##  3rd Qu.: 20271                      3rd Qu.:202505   3rd Qu.: 74.28  
##  Max.   :693099                      Max.   :219622   Max.   :100.14  
##    Fuel_Price         CPI         Unemployment     week_number   
##  Min.   :2.472   Min.   :126.1   Min.   : 3.879   Min.   : 1.00  
##  1st Qu.:2.933   1st Qu.:132.0   1st Qu.: 6.891   1st Qu.:14.00  
##  Median :3.452   Median :182.4   Median : 7.866   Median :26.00  
##  Mean   :3.361   Mean   :171.2   Mean   : 7.960   Mean   :25.83  
##  3rd Qu.:3.738   3rd Qu.:212.4   3rd Qu.: 8.567   3rd Qu.:38.00  
##  Max.   :4.468   Max.   :227.2   Max.   :14.313   Max.   :52.00

Separar año, mes y dia

bd5<- bd4
bd5<- bd5 %>%
  dplyr::mutate(year = lubridate::year(Date),
                                  month = lubridate::month(Date),
                                   day = lubridate::day(Date))

Generar regresion

regresion <- lm (Weekly_Sales ~ Store + Dept + IsHoliday + Type + Size + Temperature + Fuel_Price + CPI + Unemployment + week_number + year + month + day, data=bd5)
summary (regresion)
## 
## Call:
## lm(formula = Weekly_Sales ~ Store + Dept + IsHoliday + Type + 
##     Size + Temperature + Fuel_Price + CPI + Unemployment + week_number + 
##     year + month + day, data = bd5)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -34331 -12895  -5852   5626 671540 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)    1.110e+06  2.999e+05   3.701 0.000214 ***
## Store         -1.426e+02  3.087e+00 -46.198  < 2e-16 ***
## Dept           1.108e+02  1.097e+00 101.013  < 2e-16 ***
## IsHolidayTRUE  8.511e+02  1.391e+02   6.119 9.45e-10 ***
## TypeB         -3.133e+02  1.078e+02  -2.908 0.003642 ** 
## TypeC          5.836e+03  1.840e+02  31.709  < 2e-16 ***
## Size           9.920e-02  9.584e-04 103.511  < 2e-16 ***
## Temperature    3.701e+00  2.133e+00   1.735 0.082688 .  
## Fuel_Price     4.791e+02  1.480e+02   3.237 0.001207 ** 
## CPI           -2.340e+01  9.996e-01 -23.409  < 2e-16 ***
## Unemployment  -2.538e+02  2.062e+01 -12.308  < 2e-16 ***
## week_number    7.678e+02  4.566e+02   1.682 0.092648 .  
## year          -5.485e+02  1.485e+02  -3.695 0.000220 ***
## month         -3.167e+03  1.988e+03  -1.594 0.111036    
## day           -1.281e+02  6.539e+01  -1.959 0.050115 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 21690 on 420197 degrees of freedom
## Multiple R-squared:  0.08982,    Adjusted R-squared:  0.08979 
## F-statistic:  2962 on 14 and 420197 DF,  p-value: < 2.2e-16

Construir un modelo de prediccion

datos_nuevos <- data.frame(Store=1, IsHoliday= TRUE, Dept=1, Type="A", Size= 151315, week_number =1, Temperature = 59.17, Fuel_Price = 3.524, CPI= 214.8372, Unemployment = 7.682, year =2012, month = 1, day=1)
predict(regresion,datos_nuevos)
##        1 
## 14667.94

Conclusión

Con la base de datos de Walmart se buscaba conocer los datos, analizarlos y de ese modo predecir las ventas que se obtendrían semanalmente.

Para poder hacer esa predicción, primero tuvimos que hacer una limpieza en la base de datos. Con nuestro analisis nos dimos cuenta que dos de las bases de datos tenían la fecha como carácter. También en la base de datos “features” se veían muchos NA´s los cuales no nos sirve tenerlos en la base de datos al no aportar valor y ser un estorbo para el analisis de los datos. Y otra cosa que nos dimos cuenta es que en la base de datos “train” se tenian ventas negativas.

A partir de ese análisis de nuestras bases de datos, comenzamos la limpieza de las bases de datos.Después continuamos con la creación de una regresión, ese ese modo podíamos obtener la información necesaria para predecir las ventas semanales que permita tener una mejor toma de decisiones con datos más precisos y asertivos y pudimos darnos cuenta que se predicen 14667.94 ventas semanales, este dato analizado con distintas variables seleccionadas.

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