Introduction

In this presentation, I will demonstrate time series analysis on a dataset which provides daily climate data from January 1, 2013 to April 24, 2017 in the city of Delhi, India. The variables measured are mean temperature (celsius), humidity (g.m^-3), wind speed (kmph), and mean pressure (atm).

library(readr)
library(stringr)
library(dplyr)
library(lubridate)
library(ggplot2)
library(gt)

urlfile <- "https://raw.githubusercontent.com/kristinlussi/607ExtraCredit3/main/DailyDelhiClimateTrain.csv"

delhiClimate <- read_csv(url(urlfile), show_col_types = FALSE)

Here is a glimpse of the data:

head(delhiClimate) 

## # A tibble: 6 × 5
##   date       meantemp humidity wind_speed meanpressure
##   <date>        <dbl>    <dbl>      <dbl>        <dbl>
## 1 2013-01-01    10        84.5       0           1016.
## 2 2013-01-02     7.4      92         2.98        1018.
## 3 2013-01-03     7.17     87         4.63        1019.
## 4 2013-01-04     8.67     71.3       1.23        1017.
## 5 2013-01-05     6        86.8       3.7         1016.
## 6 2013-01-06     7        82.8       1.48        1018

Yearly Averages

Here, we will calculate the yearly averages for each year (2013 - 2016).

yearly_average <- delhiClimate %>%
  mutate(year = year(delhiClimate$date)) %>%
  group_by(year) %>%
  summarize(
    meanTemp = mean(meantemp, na.rm = TRUE),
    humidity = mean(humidity, na.rm = TRUE),
    windSpeed = mean(wind_speed, na.rm = TRUE),
    meanPressure = mean(meanpressure, na.rm = TRUE)
  ) 

# filter out year 2017
yearly_average <- yearly_average %>%
  filter(year != 2017)

yearly_average_tbl

30 Day Moving Average

Here, I calculated the 30-day moving averages for each variable using filter function from the stats package. I used 30 days here instead of 6 because we are looking at a long stretch of data (2013-2016). It is easier to see long-term trends when using a longer moving average.

# calculate 30-day moving average for mean temperature column
delhiClimate$temp_ma <- stats::filter(delhiClimate$meantemp, filter = rep(1/30, 30), sides = 2)

# calculate 30-day moving average for humidity column
delhiClimate$humidity_ma <- stats::filter(delhiClimate$humidity, filter = rep(1/30, 30), sides = 2)

# calculate 30-day moving average for wind speed column
delhiClimate$windspeed_ma <- stats::filter(delhiClimate$wind_speed, filter = rep(1/30, 30), sides = 2)

# calculate 30-day moving average for mean pressure column 
delhiClimate$pressure_ma <- stats::filter(delhiClimate$meanpressure, filter = rep(1/30, 30), sides = 2)

Time Series Plot for Mean Temperature

ggplot(data = delhiClimate) +
  geom_line(aes(x = date, y = temp_ma, color = factor(year(date)), group = year(date)), 
            na.rm = TRUE, linewidth = 1) + 
  labs(x = "Year", y = "Mean Temperature (°C)", title = "30 Day Moving Average of Mean Temperature") +
  theme_minimal() +
  theme(legend.position = "none")

As you can see from the plot, there is a similar trend from year to year, with higher temperatures in the summer months and lower temperatures in the winter months.

Time Series Plot for Humidity

ggplot(data = delhiClimate) +
  geom_line(aes(x = date, y = humidity_ma, color = factor(year(date)), group = year(date)), 
            na.rm = TRUE, linewidth = 1) +
  labs(x = "Year", y = "Humidity", title = "30 Day Moving Average of Humidity") +
  theme_minimal() +
  theme(legend.position = "none")

As you can see from the above plot, there is a slight trend in humidity levels from year to year. The humidity is lower in the spring and higher in the summer and winter months.

Time Series Plot for Wind Speed

ggplot(data = delhiClimate) +
  geom_line(aes(x = date, y = windspeed_ma, color = factor(year(date)), group = year(date)), 
            na.rm = TRUE, linewidth = 1) +  
  labs(x = "Year", y = "Wind Speed (kmph)", title = "30 Day Moving Average of Wind Speed") +
  theme_minimal() +
  theme(legend.position = "none")

From the above plot, we can see that there is a slight trend in wind speed from year-to-year. Wind speed tends to be highest in the middle of the year (June).

Time Series Plot for Mean Pressure

ggplot(data = delhiClimate) +
  geom_line(aes(x = date, y = pressure_ma, color = factor(year(date)), group = year(date)), 
            na.rm = TRUE, linewidth = 1) +  
  labs(x = "Year", y = "Mean Pressure (atm)", title = "30 Day Moving Average of Mean Pressure") +
  theme_minimal() +
  theme(legend.position = "none")

As you can see, it seems that there was a trend in the years 2013-2015, but in 2016 there was an unsual spike and decline in mean pressure levels.

Conclusion

From the analysis, we can conclude that there is a trend from year-to-year in regards to mean temperature, humidity, and wind speed. However, we cannot conclude that there is a trend from year-to-year in regards to mean pressure, since there was an unusual spike and decline in 2016 which we are not able to explain with the data provided.

Sources

Kaggle Dataset:

Title: Daily Climate Time Series Data

Author/Uploader: Sumanthvrao

URL: https://www.kaggle.com/datasets/sumanthvrao/daily-climate-time-series-data

TradeStation Help Page:

Title: What is a Trend?

Publisher/Website: TradeStation

URL: https://help.tradestation.com/10_00/eng/tradestationhelp/data/what_trend.html

Fidelity

Title: Moving Average Trading Signal - Stocks

Publisher/Website: Fidelity

URL: https://www.fidelity.com/viewpoints/active-investor/moving-averages#:~:text=A%20longer%20moving%20average%2C%20such,to%20assess%20short%2Dterm%20patterns.