Confidential - Oracle Restricted

Time Series Analysis

Understanding the Fundamentals of Time Series

Analysis

Confidential - Oracle Restricted

Confidential - Oracle Restricted

FORECASTING STOCK EXCHANGE DATA

 Price
 Yield Curve
 Exchange rates
 Volatility
https://www.bseindia.com/Indices/IndexArchiveData.html
https://www.nseindia.com/get-quotes/equity?symbol=HINDUNILVR
https://www.fbil.org.in/#/home
Foreign Exchange Dealers' Association of India

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Time series analysis

 Time series analysis is a statistical methodology appropriate for an important
class of longitudinal research designs.

 Such designs typically involve single subjects or research units (e.g.,

incidence) that are measured repeatedly at regular intervals over a large
number of observations.

 A time series analysis can help us to understand the underlying naturalistic

process, the pattern of change over time, or evaluate the effects of either a
planned or unplanned intervention

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Application of time series analysis

 Stock market analysis

 Risk analysis and evaluation of changes
 Utility studies, Budgetary analysis
 Process and quality control
 Inventory studies
 Census analysis
 Economic forecasting
 Stock market analysis

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Components of a time series

 Trend-values have a tendency to increase or decrease over time.
The Air Pollution in DELHI NCR over time. Bangalore House rentals

 Seasonal variation-similar patterns appear in corresponding seasons in

successive years.
Tomato prices .
 Cyclical variation-variation of any fixed period of time. Long-term
fluctuations that are not periodic but occur over irregular intervals.
Economic Slowdown,
 Random variation-variation that does not exhibit any fixed pattern over
time. Random variation, errors, or residuals that cannot be explained by the
trend, seasonality, or cyclic components

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Time Series Decomposition

 Additive Model:
• Y(t) = Trend + Seasonal + Irregular
• Best used when seasonal variations remain roughly constant over time.
• Multiplicative Model:
• Y(t) = Trend × Seasonal × Irregular
• Best used when seasonal variations increase or decrease over time

Confidential - Oracle Restricted

SEASONAL TREND
Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Seasonal variation

 Seasonal variation: • Seasonal variation are short term fluctuation in a time

series which occur periodically in a year:
 • This continues to repeat year after year.

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Cyclical fluctuation

 Cyclical variations are recurrent upward or downward movements in a time

series, but the period of cycle is mostly greater than a year.
 These variations are regular

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Irregular/Random Variation

 Irregular variations are fluctuation in time series that are short in duration,
erratic in nature and follow no regularity in the occurrence pattern.
 These variations are also referred what is left out in a time series after
cyclical and seasonal variations.
 Irregular fluctuations results due to the occurrence of unforeseen events
 Covid
 2008 Crisis

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Analysing time series data

 To understand the mechanism that generated the series in order to produce a

model that can be used to predict future values of the series.

 To assess the impact of some exposure on the series, after taking account of
confounding variables
 If we are concerned with the relationship between two series which show a
similar trend (e.g. a seasonal trend, or a gradual increase over time), then
the two series will be correlated, even if there is no underlying causal
relationship between them.
 For the purposes of analysis, we usually reduce our time series to a stationary
time series, in which there is no trend and cyclic variation does not increase
or decrease over time.

Confidential - Oracle Restricted

Confidential - Oracle Restricted

 If we are concerned with the relationship between two series which show a
similar trend (e.g. a seasonal trend, or a gradual increase over time), then
the two series will be correlated, even if there is no underlying causal
relationship between them. •

 For the purposes of analysis, we usually reduce our time series to a stationary
time series, in which there is no trend and cyclic variation does not increase
or decrease over time.

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Moving Average on Share Price company A

5,000.00

4,500.00

4,000.00

3,500.00

3,000.00

2,500.00

2,000.00

1,500.00

1,000.00

500.00

0.00

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Time Series Forecasting Methodologies

 ARIMA (AutoRegressive Integrated Moving Average)

 Combines autoregression (AR), differencing (I), and moving average (MA)
components.
 Exponential Smoothing
 Forecasting based on weighted averages of past observations
Model Evaluation:

 RMSE (Root Mean Squared Error), MAE (Mean Absolute Error) for
assessing forecasting accuracy.

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Data Selection for Time Series Analysis

Data Requirements:

•Frequency: Determine the frequency of data collection (e.g., daily, monthly, quarterly).
•Length: Sufficient data points (at least 50–100 data points for robust modeling).
•Relevance: Data should reflect the phenomenon you’re studying (e.g., stock prices, economic indicators).
•Data Sources:
•Government databases (e.g., Federal Reserve, World Bank)
•Private datasets (e.g., financial market data, sales data)
•Web scraping for real-time data (e.g., weather data, social media trends)

Confidential - Oracle Restricted

Confidential - Oracle Restricted

Data Exploration and Preprocessing

 Visualizing the Data:

• Time plots: To inspect trends, seasonality, and noise.
• Autocorrelation plots (ACF/PACF): To check for correlation between data points over
time.
• Handling Missing Data:
• Interpolation or forward/backward filling.
• Imputation techniques (mean imputation, median imputation).
• Detecting Outliers:
• Visual inspection (box plots, scatter plots).
• Statistical tests (Z-score, IQR).
• Transformations:
• Log transformation, differencing to stabilize variance and remove t

Confidential - Oracle Restricted

Confidential - Oracle Restricted

THANK YOU

For your Patience

Confidential - Oracle Restricted