hackathon_talento
Sergio Berdiales
6/4/2019
library(tidyverse)
library(readxl)
library(VIM)
library(GGally)Importaciones
meteo_2015 <- read_csv("sunlab-faro_meteo_2015.csv")
meteo_2016 <- read_csv("sunlab-faro_meteo_2016.csv")
prod_2015 <- read_csv("sunlab-faro-pv-2015.csv")
prod_2016 <- read_csv("sunlab-faro-pv-2016.csv")meteo_2015_2016 <- bind_rows(meteo_2015, meteo_2016)
prod_2015_2016 <- bind_rows(prod_2015, prod_2016)
rm(meteo_2015)
rm(meteo_2016)
rm(prod_2015)
rm(prod_2016)Echamos un vistazo a los datos de produccion
summary(prod_2015_2016)## Datetime A_Vertical - Voltage DC [V]
## Min. :2015-01-01 07:40:00 Min. :10.20
## 1st Qu.:2015-05-28 18:54:30 1st Qu.:26.65
## Median :2015-10-08 17:39:00 Median :27.95
## Mean :2015-12-04 18:29:34 Mean :26.28
## 3rd Qu.:2016-07-19 14:47:30 3rd Qu.:29.10
## Max. :2016-12-29 17:31:00 Max. :35.05
## NA's :13979
## A_Vertical - Current DC [A] A_Vertical - Power DC [W]
## Min. :0.000 Min. : 0.00
## 1st Qu.:0.710 1st Qu.: 15.90
## Median :1.822 Median : 50.94
## Mean :2.252 Mean : 63.56
## 3rd Qu.:3.460 3rd Qu.:100.67
## Max. :8.105 Max. :245.18
## NA's :13979 NA's :13979
## A_Optimal - Voltage DC [V] A_Optimal - Current DC [A]
## Min. :10.10 Min. : 0.000
## 1st Qu.:27.05 1st Qu.: 1.097
## Median :27.90 Median : 3.482
## Mean :27.53 Mean : 3.574
## 3rd Qu.:28.95 3rd Qu.: 5.860
## Max. :35.30 Max. :10.238
## NA's :8342 NA's :8342
## A_Optimal - Power DC [W] A_Horizontal - Voltage DC [V]
## Min. : 0.00 Min. :10.10
## 1st Qu.: 30.74 1st Qu.:26.90
## Median : 99.79 Median :27.70
## Mean :100.48 Mean :26.53
## 3rd Qu.:165.06 3rd Qu.:28.90
## Max. :307.64 Max. :35.35
## NA's :8342 NA's :9039
## A_Horizontal - Current DC [A] A_Horizontal - Power DC [W]
## Min. : 0.000 Min. : 0.00
## 1st Qu.: 1.095 1st Qu.: 27.55
## Median : 2.930 Median : 83.44
## Mean : 3.163 Mean : 87.10
## 3rd Qu.: 5.055 3rd Qu.:140.21
## Max. :10.238 Max. :304.63
## NA's :9039 NA's :9039
## A_Vertical - Temperature [ºC] A_Optimal - Temperature [ºC]
## Min. : 1.80 Min. : 2.0
## 1st Qu.:22.60 1st Qu.:21.5
## Median :30.20 Median :29.0
## Mean :29.21 Mean :28.8
## 3rd Qu.:35.80 3rd Qu.:36.0
## Max. :52.80 Max. :58.6
##
## A_Horizontal - Temperature [ºC] B_Vertical - Voltage DC [V]
## Min. : 1.40 Min. :10.30
## 1st Qu.:20.70 1st Qu.:26.45
## Median :28.30 Median :27.00
## Mean :28.59 Mean :25.48
## 3rd Qu.:36.30 3rd Qu.:27.60
## Max. :64.80 Max. :34.65
## NA's :11201
## B_Vertical - Current DC [A] B_Vertical - Power DC [W]
## Min. :0.000 Min. : 0.00
## 1st Qu.:0.678 1st Qu.: 14.95
## Median :1.847 Median : 49.89
## Mean :2.382 Mean : 64.48
## 3rd Qu.:3.695 3rd Qu.:102.69
## Max. :9.078 Max. :255.99
## NA's :11201 NA's :11201
## B_Optimal - Voltage DC [V] B_Optimal - Current DC [A]
## Min. :10.35 Min. : 0.000
## 1st Qu.:26.70 1st Qu.: 1.083
## Median :27.35 Median : 3.438
## Mean :26.31 Mean : 3.654
## 3rd Qu.:28.00 3rd Qu.: 6.250
## Max. :35.15 Max. :10.238
## NA's :6730 NA's :6730
## B_Optimal - Power DC [W] B_Horizontal - Voltage DC [V]
## Min. : 0.00 Min. :10.65
## 1st Qu.: 27.08 1st Qu.:26.70
## Median : 96.81 Median :27.35
## Mean : 99.69 Mean :26.26
## 3rd Qu.:170.25 3rd Qu.:28.35
## Max. :302.01 Max. :35.10
## NA's :6730 NA's :11902
## B_Horizontal - Current DC [A] B_Horizontal - Power DC [W]
## Min. : 0.000 Min. : 0.00
## 1st Qu.: 1.077 1st Qu.: 25.83
## Median : 2.857 Median : 81.42
## Mean : 3.120 Mean : 85.54
## 3rd Qu.: 4.980 3rd Qu.:139.04
## Max. :10.238 Max. :300.98
## NA's :11902 NA's :11902
## B_Vertical - Temperature [ºC] B_Optimal - Temperature [ºC]
## Min. : 3.20 Min. : 2.30
## 1st Qu.:23.50 1st Qu.:22.50
## Median :30.50 Median :30.90
## Mean :29.53 Mean :30.67
## 3rd Qu.:35.70 3rd Qu.:38.70
## Max. :51.10 Max. :62.00
##
## B_Horizontal - Temperature [ºC]
## Min. : 1.60
## 1st Qu.:21.10
## Median :28.30
## Mean :28.59
## 3rd Qu.:36.10
## Max. :62.10
## glimpse(prod_2015_2016)## Observations: 420,507
## Variables: 25
## $ Datetime <dttm> 2015-01-09 12:52:00, 2015-0...
## $ `A_Vertical - Voltage DC [V]` <dbl> 29.35, 29.10, 29.10, 29.35, ...
## $ `A_Vertical - Current DC [A]` <dbl> 6.4125, 6.3975, 6.3325, 6.41...
## $ `A_Vertical - Power DC [W]` <dbl> 188.206875, 186.167250, 184....
## $ `A_Optimal - Voltage DC [V]` <dbl> 28.75, 28.45, 28.45, 28.75, ...
## $ `A_Optimal - Current DC [A]` <dbl> 7.1700, 7.1925, 7.0925, 7.21...
## $ `A_Optimal - Power DC [W]` <dbl> 206.13750, 204.62662, 201.78...
## $ `A_Horizontal - Voltage DC [V]` <dbl> 29.25, 29.50, 29.45, 29.50, ...
## $ `A_Horizontal - Current DC [A]` <dbl> 4.3225, 4.2525, 4.1825, 4.30...
## $ `A_Horizontal - Power DC [W]` <dbl> 126.43313, 125.44875, 123.17...
## $ `A_Vertical - Temperature [ºC]` <dbl> 32.6, 32.4, 33.2, 32.2, 32.8...
## $ `A_Optimal - Temperature [ºC]` <dbl> 28.8, 28.6, 29.4, 28.2, 27.8...
## $ `A_Horizontal - Temperature [ºC]` <dbl> 26.5, 26.1, 26.8, 26.4, 26.1...
## $ `B_Vertical - Voltage DC [V]` <dbl> 27.55, 27.60, 27.40, 27.40, ...
## $ `B_Vertical - Current DC [A]` <dbl> 7.0925, 7.0475, 7.0100, 7.14...
## $ `B_Vertical - Power DC [W]` <dbl> 195.39837, 194.51100, 192.07...
## $ `B_Optimal - Voltage DC [V]` <dbl> 27.85, 28.05, 27.85, 28.05, ...
## $ `B_Optimal - Current DC [A]` <dbl> 7.2225, 7.1125, 7.0450, 7.17...
## $ `B_Optimal - Power DC [W]` <dbl> 201.14663, 199.50563, 196.20...
## $ `B_Horizontal - Voltage DC [V]` <dbl> 29.05, 29.05, 28.75, 28.95, ...
## $ `B_Horizontal - Current DC [A]` <dbl> 4.1775, 4.1275, 4.1125, 4.19...
## $ `B_Horizontal - Power DC [W]` <dbl> 121.356375, 119.903875, 118....
## $ `B_Vertical - Temperature [ºC]` <dbl> 31.2, 31.5, 31.9, 31.7, 32.2...
## $ `B_Optimal - Temperature [ºC]` <dbl> 30.2, 30.0, 30.9, 29.9, 30.7...
## $ `B_Horizontal - Temperature [ºC]` <dbl> 25.3, 24.9, 25.7, 25.1, 25.6...prod_2015_2016_final <- prod_2015_2016 %>%
select(Datetime,
'A_Optimal - Power DC [W]',
'B_Optimal - Power DC [W]')summary(meteo_2015_2016)## Datetime Ambient Temperature [ºC]
## Min. :2015-01-01 00:00:00 Min. :-1.500e+09
## 1st Qu.:2015-05-26 11:06:30 1st Qu.: 1.400e+01
## Median :2015-10-18 23:12:00 Median : 1.800e+01
## Mean :2015-11-27 03:00:27 Mean :-5.358e+03
## 3rd Qu.:2016-05-17 18:32:30 3rd Qu.: 2.300e+01
## Max. :2016-12-31 23:00:00 Max. : 3.800e+01
##
## Global Radiation [W/m2] Diffuse Radiation [W/m2] Ultraviolet [W/m2]
## Min. : 0.7285 Min. : 1.6 Min. : 0.4954
## 1st Qu.: 1.5693 1st Qu.: 1.7 1st Qu.: 0.5106
## Median : 13.0063 Median : 16.5 Median : 1.5008
## Mean : 237.1203 Mean : 68.2 Mean :14.5440
## 3rd Qu.: 452.7272 3rd Qu.:103.7 3rd Qu.:26.0017
## Max. :1421.3282 Max. :720.4 Max. :84.2660
## NA's :399744
## Wind Velocity [m/s] Wind Direction [º] Precipitation [mm]
## Min. :-1.500e+09 Min. : 0.0 Min. :0
## 1st Qu.: 1.000e+00 1st Qu.:123.6 1st Qu.:0
## Median : 2.000e+00 Median :235.8 Median :0
## Mean :-5.375e+03 Mean :212.8 Mean :0
## 3rd Qu.: 3.000e+00 3rd Qu.:296.7 3rd Qu.:0
## Max. : 1.200e+01 Max. :360.0 Max. :2
## NA's :776160
## Atmospheric pressure [hPa] Direct Radiation [W/m2]
## Min. : 999.1 Min. : 1.6
## 1st Qu.:1014.3 1st Qu.: 1.7
## Median :1020.4 Median : 8.6
## Mean :1019.9 Mean : 63.9
## 3rd Qu.:1025.7 3rd Qu.: 91.9
## Max. :1033.8 Max. :790.1
## NA's :776160 NA's :437159meteo_2015_2016_final <- meteo_2015_2016 %>%
select(Datetime,
'Ambient Temperature [ºC]',
'Diffuse Radiation [W/m2]',
'Ultraviolet [W/m2]',
'Global Radiation [W/m2]')Cruzamos las tablas por Datetime
tabla_final <- meteo_2015_2016_final %>%
left_join(prod_2015_2016_final) %>%
filter(`Ambient Temperature [ºC]` <1000)summary(tabla_final) ## Datetime Ambient Temperature [ºC]
## Min. :2015-01-01 00:00:00 Min. :-1.500e+09
## 1st Qu.:2015-05-26 11:06:30 1st Qu.: 1.400e+01
## Median :2015-10-18 23:12:00 Median : 1.800e+01
## Mean :2015-11-27 03:00:27 Mean :-5.358e+03
## 3rd Qu.:2016-05-17 18:32:30 3rd Qu.: 2.300e+01
## Max. :2016-12-31 23:00:00 Max. : 3.800e+01
##
## Diffuse Radiation [W/m2] Ultraviolet [W/m2] Global Radiation [W/m2]
## Min. : 1.6 Min. : 0.4954 Min. : 0.7285
## 1st Qu.: 1.7 1st Qu.: 0.5106 1st Qu.: 1.5693
## Median : 16.5 Median : 1.5008 Median : 13.0063
## Mean : 68.2 Mean :14.5440 Mean : 237.1203
## 3rd Qu.:103.7 3rd Qu.:26.0017 3rd Qu.: 452.7272
## Max. :720.4 Max. :84.2660 Max. :1421.3282
## NA's :399744
## A_Optimal - Power DC [W] B_Optimal - Power DC [W]
## Min. : 0.0 Min. : 0.0
## 1st Qu.: 29.8 1st Qu.: 26.1
## Median : 97.8 Median : 94.9
## Mean : 99.4 Mean : 98.7
## 3rd Qu.:163.1 3rd Qu.:169.3
## Max. :307.6 Max. :302.0
## NA's :479354 NA's :477901write_csv(tabla_final, "meteo_prod_2015_2016.csv")Preparar datos de test
Importaciones
test_meteo_2017 <- read_csv("test-sunlab-meteo-2017.csv")
test_prod_2017 <- read_csv("test-sunlab-pv-2017.csv")Cruzamos las tablas por Datetime
tabla_final_test <- test_meteo_2017 %>%
left_join(test_prod_2017, by = "Datetime") %>%
select(`A_Optimal - Power DC [W]`,
`Global Radiation [W/m2]`,
`Ultraviolet [W/m2]`)write_csv(tabla_final_test, "test_meteo_prod_2017.csv")Echamos un vistazo a la completitud de los datos
aggr(tabla_final)
Panel A
tabla_final_A <- tabla_final %>%
select(-`B_Optimal - Power DC [W]`) %>%
na.omit()
modelo_A_1=lm(`A_Optimal - Power DC [W]` ~ `Global Radiation [W/m2]` + `Ultraviolet [W/m2]` , data = tabla_final_A)Echamos un vistazo al modelo
modelo_A_1##
## Call:
## lm(formula = `A_Optimal - Power DC [W]` ~ `Global Radiation [W/m2]` +
## `Ultraviolet [W/m2]`, data = tabla_final_A)
##
## Coefficients:
## (Intercept) `Global Radiation [W/m2]`
## 4.8996 0.2692
## `Ultraviolet [W/m2]`
## -0.9980summary(modelo_A_1)##
## Call:
## lm(formula = `A_Optimal - Power DC [W]` ~ `Global Radiation [W/m2]` +
## `Ultraviolet [W/m2]`, data = tabla_final_A)
##
## Residuals:
## Min 1Q Median 3Q Max
## -217.181 -18.974 -5.129 13.335 228.712
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.899607 0.115663 42.36 <2e-16 ***
## `Global Radiation [W/m2]` 0.269219 0.001342 200.54 <2e-16 ***
## `Ultraviolet [W/m2]` -0.997968 0.022165 -45.02 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 30.48 on 221114 degrees of freedom
## Multiple R-squared: 0.8266, Adjusted R-squared: 0.8266
## F-statistic: 5.269e+05 on 2 and 221114 DF, p-value: < 2.2e-16fitted_values <- modelo_A_1$fitted.values
real_values <- tabla_final_A$`A_Optimal - Power DC [W]`tabla_final_test_A <- tabla_final_test %>%
select(`A_Optimal - Power DC [W]`,
`Global Radiation [W/m2]`,
`Ultraviolet [W/m2]`)Hacemos la prediccion para los paneles A
prediccion_test <- predict(modelo_A_1, tabla_final_test_A)prediccion_test <- as.data.frame(prediccion_test)
write_csv(prediccion_test, "predicciones_panel_A.csv")tabla_test_predicciones_A <- bind_cols(tabla_final_test_A, prediccion_test) %>%
na.omit()tabla_test_predicciones_A_1 <- tabla_test_predicciones_A %>%
mutate(diff = abs(`A_Optimal - Power DC [W]` - prediccion_test))mean(tabla_test_predicciones_A_1$diff, na.rm = TRUE)## [1] NaN