Predictive Analysis in data science is the branch of advanced analytics which is used to make predictions about unknown future events. Predictive analytics uses numerous methods from artificial intelligence, machine learning, modeling, statistics, and data mining to evaluate current details to make predictions with regards to the Future.
Hello everyone, I hope everyone is doing well. So, we have already gone through :
1. What is data science and why do we need this now?
2. What is Data Collection in Data Science?
3. What is a Descriptive Analysis?
So let’s begin where we left. I talked about data collection and descriptive analysis in the previous article. So today I will be sharing:
1. What is Predictive Analysis?
2. Examples of Predictive Analysis
Predictive analysis is one of my favorite ones as this is being used everywhere. So If I have to compare in between the descriptive and predictive analysis, they are somehow interrelated to each other and I can explain why?
So if you remember what a descriptive analysis is, it’s basically wherein we analyze what happened in the past? What could be the reason that it happened?
Most likely it’s a negative thing that has happened in the past. This is the reason you have been asked to analyze the data and information. So once you are done with the descriptive analysis, then your clients say please do predictive analysis for the next 6-8 months or probably 1 year and see how my stocks will perform or my business will do basis the historical data that you have.
Using Predictive Analysis, you try to predict something. So predictive analysis is the branch of advanced analytics which is used to make predictions about unknown future events. Predictive analytics uses numerous methods from artificial intelligence, machine learning, modeling, statistics, and data mining to evaluate current details to make predictions with regards to the Future.
Examples: To start with let’s talk about stocks. There are several stocks listed in NSE which are the National Stock Exchange and BSE which is Bombay Stock Exchange. You have a client who has bought several stocks. He reaches on to you saying that I have these 20 stocks with me, and I have invested over 1cr or 1.5cr in these stocks. I want to know if :
1. It is the right decision to keep these stocks with me for the next 5 years or the next 3 years or the next 2 years.
2. Can you do predictive analysis for me so that I am able to take that decision?
This is when your job starts. You will pick up 5 years’ performance data of those stocks. Usually, clients do provide the data. However, in some cases, we need to procure the relevant date ourselves through various free as well as paid sources.
Once you have those 5 years of historical information about the performance of those stocks, you start by applying techniques like data mining, statistics, modeling, machine learning, artificial intelligence. You apply those techniques and try to predict the future for the relevant stocks that are owned by your client.
Guys trust me; predictive analysis is the most famous one in the market. Every other organization is doing that because they want to know where do they stand in the coming 6-8-11 months. In Fact, MNCs generally create a 5 years roadmap.
So coming back to our example about the 5 years performance of those stocks which is being kept by your client. You will start by taking into account several factors that can have an effect on those stocks. They could be :
1. How parent companies of these stocks are doing in the market?
2. How that particular stock sector is doing?
3. Are there any expansion plans or merger coming up in the near future?
Just to make you aware, there is no defined time deadline for the whole process. Even after getting a higher degree of accuracy in your predictions, you have to keep applying techniques, understand, see how it goes and then share the analysis outcome with your clients. You can suggest your client that out of these 20 stocks you can keep these 10 stocks for the next 5 years which will grow by 40 to 60 percent, but the rest of the 10 stocks may not perform well. So it is my suggestion to sell out those 10 stocks and keep the rest.
Now, it is up to the client if they want to follow your suggestions. This is how predictive analytics works. Your level of predictions depends on the accuracy you have achieved. If your accuracy is 70%, 80% or 90 %, you have done a great job. 90 percent is the rarest of a rare case, otherwise, 40 or 50 or 60 is usually what you want to expect at the start. In case, you hit an accuracy level of 80-90 percent, that is great job done. Your client will close his eyes and will buy your analysis right away.
Predictive analysis applications : Predictive analysis can be applied everywhere. Look at weather forecasting. You are able to find in Google what is going to be the weather conditions in Delhi for the following five days or perhaps what’ll be the weather in the following two weeks. The moment you type in, you get the weather information and most likely that will be the right information.
How does Google do that?
Well, the Google team has historical data of weather with them. Basis of the data, they can predict what will be the weather in the future. Of course, you have satellite images as well how’s the weather across the globe but usually, your weather forecast is done basis the historical data. So if it will be raining in June or it will be very hot in the month of May, it can be predicted via predictive analysis.
Nowadays, companies are using predictive analytics to solve their hiring pain areas. Let’s say, an organization is trying to hire few people but they are not sure if they should go ahead with that particular hiring or not. So they reach out to their data analytics team to ask them if they will have that sort of business wherein they will need more people.
But please make a note that predictive analysis is incomplete without descriptive analysis. Because it is imperative that you should know what happened in the past which will help you draw patterns among the various data points. It is only that once you have that analysis done, you can start predicting the future.
There are a plethora of live business examples wherein predictive analysis has done wonders for an organization. One of its recent examples is cryptocurrency. You must have heard of Bitcoins. So there was an analysis done by a few data scientists back in the US that how cryptocurrency-based bitcoins will perform in the next 2 years. Their analyses were almost accurate. They said that cryptocurrency is going to grew by 150 to 100 percent and that is what happened in early 2018. The value of bitcoins grew from 15 dollars to 15000 dollars in a very short span of time. People became millionaires overnight. So that is the impact we are talking about of data analysis or rather predictive analysis.