Making Numerical Predictions For Time Series Data - Part 1/3
Master Basics To Advance Tools For Predictive Analytics On Time Series Data Using Descriptive Statistics Moving Averages Regressions Machine Learning Neural Networks!
- Self-paced with Life Time Access
- Certificate on Completion
- Access on Android and iOS App
Course Preview Video
-
Author
-
Categories
-
Duration
Learn All About Predictive Analytics On Time Series Using Advanced Methods
Predictive analytics is the branch of the advanced analytics which is used to make predictions about unknown future events. Predictive analytics uses many techniques from data mining, statistics, modelling, machine learning, and artificial intelligence to analyse current data to make predictions about future.
One class of Predictive Analytics is to make prediction on Time Series Data. Studying historical data, collected over a period of time, can help in building models using which future can be predicted. For example, from historical data on Temperatures in a City, we can make decent predictions of what the Temperature could be in a future date. Or for that matter, from data collected over a reasonably long period of time regarding various life style aspects of a Diabetic patient, we can predict what should be the volume of Insulin to inject on a given date in future. One example to consider from the Business world could be to predict the Volume of In-Roamers in a Telecom Network in any given period of time in the future from the historical details of In-Roamers in the Network.
The applications are just innumerable as these are applicable in every sphere of business and life.
In this course, we go through various aspects of building Predictive Analytics Models. We start with simple techniques and gradually study very advanced and contemporary techniques. We cover using Descriptive Statistics, Moving Averages, Regressions, Machine Learning and Neural Networks.
This course is a series of 3 parts.
- In Part 1, we use Excel to make Numerical Predictions from Time Series Data.
We start by using Excel for 2 reasons.
- Excel is easy use and thus we can understand complex concepts through exercises that are easy to replicate and thus become easy to understand.
- Excel is expected to be available with everyone taking this course.
- In Part 2, we use R Programming to make Numerical Predictions from Time Series Data.
- In Part 3, we use Python Programming to make Numerical Predictions from Time Series Data.
The course uses simple data sets to explain the concepts and the theory aspects. As we go through the various techniques, we compare the various techniques. We also understand the circumstances where a particular technique should be applied. We will also use some publicly available data sets to apply the techniques that we will discuss in the course.
From time to time, we will add bonus videos of our real time work on industrial data on which we will apply the Predictive Analytics techniques to create Models for making predictions.
- Basic Knowledge of Statistics
- Basic Knowledge of Algebra
- Basic Knowledge of Logarithm
- Basic Knowledge of Excel
- Predicting using Descriptive Statistics, Moving Averages, Centred Moving Averages, Weighted Moving Averages
- Predicting using Linear Regression
- Predicting using Exponential Regression
- Predicting using Power Regression
- Predicting using Logarithmic Regression
- Predicting using Polynomial Regression
- Using Excel to make Predictions
- Using Data Analysis Tool Pak from Excel
- Using LINEST(), LOGEST(), GROWTH(), TREND() functions in Excel
Generally, it is seen that forecasting involves studying the behaviour of a characteristic over time and examining data for a pattern. The forecasts are made by assuming that, in future, the characteristic will continue to behave according to the same pattern.
A Time Series is a collection of observations made sequentially over a period of time.
Descriptive statistics are brief descriptive coefficients that summarize a given data set, which can be either a representation of the entire or a sample of a population. Descriptive statistics are broken down into measures of central tendency and measures of variability (spread).
A simple, but widely used, strategy to predict future demand is to use central tendencies of past data to be used as the future demand.
Moving Averages Prediction is another simple, but powerful, way to predict future values based on historic data. Moving Averages takes in to consideration recency of data.
A more popular variation of Moving Averages is Centred Moving Averages. This video discussed using Centred Moving Averages for Predicting Future Value.
Weighted Moving Averages is a more powerful tool for predicting future values as it has mechanism to give more priority to factors like RECENCY of data.
One of the methods for finding confidence in the predictions made using Moving Averages is by determining and interpreting the Standard Deviation of the Predictions.
In this video, we discuss technique for predicting future values when the Time Series Data has Seasonality Component.
Linear regression is used to predict the value of a continuous variable Y based on one or more input predictor variables X. The aim is to establish a mathematical formula between the the response variable (Y) and the predictor variables (Xs). You can use this formula to predict Y, when only X values are known.
This video demonstrates the method to use Linear Regression for making Predictions as discussed in the previous video.
This video explains how we can conduct Linear Regression using LINEST() function of Excel for 1 Dependent Variable.
When we have a Level-Level Regression, we can use the TREND() function in Excel to predict future values. TREND() is an Array-Function in Excel.
Data Analysis Tool Kit of Excel makes it very easy to conduct Regression Analysis. However, it is very vital to understand the output produced by it.
This video discusses in details the process for conducting Linear Regression Analysis with 1 independent variable using Data Analysis Tool Kit of Excel.
In this video, we discuss Linear Regression when we have more than 1 Independent Variable.
In this video, we discuss another technique for making predictions on Time Series Data, i.e. Exponential Regression using Linear Model.
The Exponential Regression using a Linear Model suffers from the shortcoming that it doesn’t actually minimise the sum of the squares of the deviations. We now show how to use Solver to create a better, nonlinear, regression model.
In this video, I discuss how we can conduct Exponential Regression using Excel Function LOGEST().
I also discuss how we can make predictions using Excel function GROWTH().
We round up our discussion on Exponential Regression with discussion on how to conduct Exponential Regression when we have to deal with more than 1 Independent Variables. We will use the LOGEST() and GROWTH() functions. We will see how we can use Data Analysis Toolkit.
In this video we discuss Power Regression when we are dealing with 1 Independent Variable. Power Regression is also called Log-Log Regression.
When the data demonstrate a trend that it grows or decays rapidly in the beginning and slows down later, we can use Logarithmic Regression.
In this video, we discuss all about Logarithmic Regression.
In this video, we discuss a Non-Linear Regression Model i.e. Quadratic Regression.
In this video, we discuss another non-linear regression model i.e. Polynomial Regression.
Before we can settle down to a Model which we can use for making reliable predictions, we need to go through a process of experimenting with a lot of alternatives and studying their performance. This video demonstrates the process. The video does not provide the ultimate process or does not encompass all the alternatives of Modelling. This video just illustrates the process of making selection of a model and some considerations that can be studied in the process.
For the illustration, we only create models with one Independent variable. Needless to say, we must examine many more options involving many more variables. We will examine more options in the later parts of this series.
Outliers may be genuine data or may be erroneous data. In either case, it is essential to identify them and form strategy for dealing with them.
In this video, I discuss Normal Distribution. This video does not get into the Calculus involved. This has been done so that everyone can find this video easy to follow and make use of it.
When we are dealing with very large Data Sets, it is difficult to establish the Population Mean. So, we need to estimate the Population Mean. The Interval of Confidence with which we estimate the Population Mean is known as the Confidence Interval.
