Time Series Analysis using Python

Time series analysis means analyzing and finding patterns in a time series dataset. A time-series dataset is a sequence of data collected over an interval of time. Stock price data, monthly sales data, daily rainfall data, hourly website traffic data are some examples of time-series data that you will get to solve business problems as a data scientist. So if you want to learn Time Series Analysis, this article is for you. In this article, I will take you through the task of Time Series Analysis using Python.

Time Series Analysis

Whenever you are analyzing a dataset recorded over a time interval, you are doing Time Series Analysis. The time interval of a time series data can be weekly, monthly, daily, or even hourly time intervals, but the process of analyzing your data will remain the same in most of the problems.

At the end of this article, you will learn to do Time Series Analysis using Python. I will be using the plotly library in Python here as it is easy to analyze data with plotly because of less code and interactive results. I will recommend using a Jupyter Notebook or Google Colaboratory for Time series analysis instead of using a code editor or an IDE like VS Code or PyCharm.

Time Series Analysis using Python

Let’s start the task of Time Series Analysis using Python by importing the necessary Python libraries and a time series dataset:

import pandas as pd
import yfinance as yf
import datetime
from datetime import date, timedelta
today = date.today()

d1 = today.strftime("%Y-%m-%d")
end_date = d1
d2 = date.today() - timedelta(days=720)
d2 = d2.strftime("%Y-%m-%d")
start_date = d2

data = yf.download('AAPL', 
                      start=start_date, 
                      end=end_date, 
                      progress=False)
print(data.head())

                 Open       High        Low      Close  Adj Close     Volume
Date                                                                        
2020-01-28  78.150002  79.599998  78.047501  79.422501  78.260017  162234000
2020-01-29  81.112503  81.962502  80.345001  81.084999  79.898186  216229200
2020-01-30  80.135002  81.022499  79.687500  80.967499  79.782402  126743200
2020-01-31  80.232498  80.669998  77.072502  77.377502  76.244957  199588400
2020-02-03  76.074997  78.372498  75.555000  77.165001  76.035568  173788400

In the above code, I have used the yfinance API to extract the latest stock price data. You can learn more about it from here. Now let’s visualize a line plot to see the trends in stock prices of Apple:

import plotly.express as px
figure = px.line(data, x = data.index, 
                 y = "Close", 
                 title = "Time Series Analysis (Line Plot)")
figure.show()

A line plot is one of the best visualization tools while working on Time series analysis. In the above code, I am visualizing the trends in the close prices of Apple. If you place the cursor on the line, you will see the Close price on the exact date of the data point on which your cursor is.

Now let’s visualize a candlestick chart to see the trends in the open, high, low, and close prices of Apple:

import plotly.graph_objects as go
figure = go.Figure(data=[go.Candlestick(x = data.index,
                                        open = data["Open"], 
                                        high = data["High"],
                                        low = data["Low"], 
                                        close = data["Close"])])
figure.update_layout(title = "Time Series Analysis (Candlestick Chart)", 
                     xaxis_rangeslider_visible = False)
figure.show()

A candlestick chart is always helpful in the time series analysis of a financial instrument. If you place the cursor on any point in the above candlestick chart, you will see all the prices of Apple (open, high, low, and close) on the date where your cursor is. The red lines of this chart indicate a fall in prices, and the green lines indicate an increase in prices.

Now let’s visualize a bar plot to visualize the trends of close prices over the period:

figure = px.bar(data, x = data.index, 
                y = "Close", 
                title = "Time Series Analysis (Bar Plot)" )
figure.show()

The bar plot above shows an increase in stock prices in the long term scenario. The line chart and candlestick chart show you increase and decrease of the price, but if you want to see the price increase and decrease in the long term, you should always prefer a bar chart.

If you want to analyze stock prices between the period of two specific dates, then below is how you can do it:

figure = px.line(data, x = data.index, 
                 y = 'Close', 
                 range_x = ['2021-07-01','2021-12-31'], 
                 title = "Time Series Analysis (Custom Date Range)")
figure.show()

One of the best ways to analyze a time series data is to create an interactive visualization where you can manually select the time interval in the output visualization itself. One way to do it is to add a slider below your visualization and buttons to control time intervals above your visualization. Below is how you can create an interactive candlestick chart where you can select time intervals in the output itself:

figure = go.Figure(data = [go.Candlestick(x = data.index,
                                        open = data["Open"], 
                                        high = data["High"],
                                        low = data["Low"], 
                                        close = data["Close"])])
figure.update_layout(title = "Time Series Analysis (Candlestick Chart with Buttons and Slider)")

figure.update_xaxes(
    rangeslider_visible = True,
    rangeselector = dict(
        buttons = list([
            dict(count = 1, label = "1m", step = "month", stepmode = "backward"),
            dict(count = 6, label = "6m", step = "month", stepmode = "backward"),
            dict(count = 1, label = "YTD", step = "year", stepmode = "todate"),
            dict(count = 1, label = "1y", step = "year", stepmode = "backward"),
            dict(step = "all")
        ])
    )
)
figure.show()

So this is how you can perform time series analysis using Python.

Summary

I hope you now have understood how to do Time Series Analysis using Python and all the visualizations you can use for Time Series Analysis. A time-series dataset is a sequence of data collected over an interval of time. Time series analysis means analyzing and finding patterns in a time series dataset. The time interval of a time series data can be weekly, monthly, daily, or even hourly time intervals. I hope you liked this article on Time Series Analysis using Python. Feel free to ask valuable questions in the comments section below.