Climate Change Prediction using Time Series | Python Projects | Edureka | DS Rewind - 5

Key Takeaways

good morning good afternoon and good evening guys based on the time zones you all are coming from so guys before we start with the session can you all please give me a quick confirmation if you all can see my screen and hear me louding c as well great thank you for confirming everyone so my name is NJ Kya and have been working in this IT industry for more than 13 years now but before we proceed further let me quickly introduce our Eda masterclass Community with you all as well so this community of master classes was started back in 2019 and since then we have been closing into almost 32,000 members so far and in these master classes we have been conducting multiple webinars on different topics including blockchain iot artificial intelligence machine learning big data and multiple front end and backend development Technologies and the best part about these webinars are they are absolutely free of cost so there are no charges involved here and these webinars are a great are a really great platform for anyone who is looking to get into this industry vertical by learning the technology that they are interested in as a part of our discussion we are going to discuss on the climate change and how exactly that is structured so that is what we are going to focus on as we proceed further so first of all we are going to talk about the principles of the time series analysis what exactly it is and then we are going to proceed further into the handson so first of all here we are going to look at what exactly is time series data what is time series analysis time series use cases about the project why does stationarity matters and environment and tools for the project including the hon so if you talk about time stre data then what exactly it is it is a collection of observations of those obtain through repeated measurements over time and it also referred to as time stamp data is a sequence of data pointers index in time model so for example we can see here we have a data for a website visitors like for example how many visitors are there for a month and then in the park and what exactly is the temperature the average temperature recorded here so we can see here we have the number of visitors per month then we have the timeline and then we have the average temperature recorded for the yellow stent Park as a part of a Time series based data so what exactly a Time series analysis is so if we talk about the antire analysis and time series analysis is basically a statistical technique that deals with time series data or the train analysis so the time stre data means that data in is in a series of particular periods of intervals so they are going to be multiple use cases for time series that we are going to talk about as we proceed further just a moment so in terms of the use cases so there are multiple use cases for time series like we have a financial services for weather analysis for network data analysis for healthcare analysis and so on so now some may have a question what exactly makes it special because see time series is a collection of data pointers collected at constant time intervals and these are analyzed to determine the long-term Trend so as to forecast the future of or perform some other form of analysis but what exactly makes a TS different from let's say a regular aggression problem so there are two things so time series is time dependent so the basic Assumption of linear regression model that the observ observations are independent doesn't hold in this case and along with an increasing or decreasing Trend most TS against the time series have some form of seasonality trends that is variation specific to a particular time frame for for example we can see the sales of a ven jacket over time you will invariably find higher sales in winter seasons and because of the inherent properties of the T of the time series there are various steps involved in analyzing it and they are and they are again they are going to be discussed and how exactly we can work with and we can work with the entire analysis based on python so we have multiple use cases for example let's say here we are talking about stock market data of 2020 from kagle so kagle is one of the primary data sources where we can get data for almost every kind and some are available from the official apis from the companies itself so basically here we have the access to multiple data sets which can be from the a we can find data set for Aviation for banking for e-commerce for for the normal analysis based on any Tech any entertainment industry as well so these all things can be defined all right so we have classification so we have data set for almost every industry for example for coid for economics for economics we have the country region and world GDP we have World happiness report Supermarket sales we have for education sectors Healthcare data we have for different image we can say image data for visualizations for NLP for arts and entertainment so we have all the data set available and we can go ahead and explore that now this is a good repository where we can find as you can see almost 80,000 data sets are available for different use cases that we can make use of so here we are going to talk about stationary so first of all let's talk about why does stationary matters so why does stationary matters here so basically most of the time most time Ser models assume that each point is independent of of one another and the statistical properties of data should not change over time and stationary helps us better identify the driving factors as well like we have the stationary time time series and then we have a non-stationary Time series and the same data set is being used in these two different graphs and then we have environments and tools for the project so basically here we do need to have the access to dpai pandas M daytime and stats model so these are the Li is required here and we are going to work on Python programming language so now as a part of the handson we are going to work on the same file that we have the access to and let me just guide you how we can work with it so this is a notebook file that we have currently opened up so here we have to first of all import bandas as PD so first of all we are working on collab if you're comfortable with your own local jupyter notebook file then we can make use of jupyter notebook or we can make use of the other notebook file as for the requirement we can do that and for now we are going to make use of uh collab so collab doesn't have any kind of Hardware requirement in case we are having in case we don't have the access to a highend system or we are experiencing a low performance in our own system then we can gohe and make use of cab it's like an online notebook available where we can code on top of python so we can Define the libraries that we are going to make use of so first of all we have to import pandas as Speedy then from daytime we are going to import daytime and and time date then we are going to import mattli which is basically used for visualization part as PLT then dumpy as NP and then we have stats model to work on the statistical models here and then cbon which is basically an advanced visualization tool offer just like we have mlet and then we want to use in line so this line is being added here because we don't want to see the cfts being opened up in in a different window we want them to be showcased within the notebook itself that we are going to work with and then we are going to import on warnings because there are sometimes some warnings are going to be shown based on the data set and method that we are going to make use of now we don't want that so we can simply go ahead and import warnings all right so first of all before we start we have to load these libraries and then we have to go ahead and run it so for running the statement we can simply click on play and this is going to run the statement for us if if we are going to work on a you can say a visualization heavy application then there we can make use of the GPU based instance here all right so basically in case we want to change the runtime we can simply Define change runtime as suppose here we want to focus on GPU based instance in case we are going to work on multiple visualization tools then we can make use of GP based C again as you can see here now again the computer engine is currently being refreshed we can execute this again so all the major libraries required have been Creed all right so once we have done importing these then we are going to work on reading the CSV file if we have a global and temperatures by state. CSV so this is basically a data that we have the access to so in case you are going to work on collab so here we can simply goad and import the data set so here we have folder by the name of Workshop files and within Workshop files we are going to upload the temperature the global land temperature file that we have the access to because we are going to work with collab right so for colab we have to make sure that we do add the Drive Link then only we can connect if in case we are doing this locally then there's no need to connect uh we there's no need to upload our data on Google Drive we can Define simply the local path but again before we can work with it we also have to mount our Google drive with this notebook file in case you want to work on Google Drive as well so here we can click on Mount so basically a code is going to be inserted here which will say from google. cab we have we are going to import drive so first of all we have to authenticate that yes we want to use our Google drive from this account we have to allow it and once we verify it we have to get the authorization code and this code is what we have to enter here in case we are going to do this locally then there's no need of mounting this in case you want to work on Google Co we can say Kola drive then only we have to use it all right so here we can go ahead and as you can see here the drive has been mounted so now we can Define the path so our content is available under content drive drive and then under this we have my drive and under this we have a folder so we have a folder by the name of Workshop files work Workshop underscore files and then under Workshop files we have the file name by global land change so this is a file name that we have to enter AS Global lter Bri state by state. CSV so this is a the file PA that we have to Define all right so so we are going to make use of DF so PD as in pandas so P using pandas we are going to Simply read the content of this file so we can run the statement here as well we can run it and you can see the statement is currently being executed and now if you want to show the head do that means the first five rows available in this feet and here we can simply run the statement here and here we have date average temperature average temperature uncertainty and then we have state and then we have the country all right so now if you want to see the types here we can simply run the DF types so you can see here we have the objects and then we have the column and then we have the data type defined for these different columns all right and then we can go ahead and print the shape that means how much of data we have and then we can simply print if it is null then what should be the sum of the values and then we are simply going to represent the first five rows in terms of the first 100 rows being being returned as a response and now we are going to define the name so here we are going to rename certain columns as D to date average temperature to average temperature average temperature uncertainty to to confidence interval in confidence interval temperature to make sure labels are more aligned to what we are trying to achieve here and then once we have changed this we can see now the this was the ear heading right The Ear Main column and again here we have the change heading and now we are going to make use of the same date to dat date time so we we are going to convert this call into into day time format itself and then we are going to set the date and then we simply going to print the DF index so next we are going to Simply Define the different temperature changes so we have we have latest differences in terms of the countries Z have tempature we have a group P by country and then we are going to Simply find the average and then we are going to S the values based on average temperature as you can see the lowest has been for Canada then we have Russia us China Australia India and Brazil in terms of average temperature being Define and then we can simply plot this on a graph as well where we have the figure size as 94 and then we are going to Simply import as 9 forn Valu so again here we Define a simple Plot show so this entire graph is going to be plotted by using the same plot library that we have twed but again before that we can do that we have to ensure that we do run all the other data frames as well because again and we have to make sure that we do create the data frames we do go ahead and import all the data types as well then only we would be able to make use of it all right so here we are simply going to create a new data frame where we going def find the latest data frame as well suppose we want to create a temporary data frame from from 9802 2013 then we can simply create a new data frame out of the existing data frame that we already have the access to all right and then we are going to create the latest dat difference so in terms of dat we are going to define the country and the average temperature based on the new data pointers that we have as average temperature so this is going to be a new view that has been defined and then we are going to create a new graph from the same lat latest as you can see this is the latest gra that has been created and then we are going to Simply work on resampling as well so after resampling we have simply resample based on the parameters as a and then we are simply going to Simply Define the resample parameters and then we are going to plot the resample our plot by using the M plot Library we are defined the title as this one figure size and then we are going to plot temperature and here and we can see it has been continuously on a rise only it has been increasing on a per year basis and then we have the other components now let's suppose here we are looking to compare the changes in 50 years slots for example we want to see how the change has been in last 50 years so for that we are going to compare the timeline in terms of this time three Anis so first of all we have to use a resample data frame and then from that we are going to exponentially V find the Ved mean and then we are going to roll standard deviation and then we are going to create a subplots next to each other and then we are simply going to to create two different graphs here temperature graphs with rolling mean and exponentially weighted mean as well and then we are going to create a temperature graph with rolling SD where we get to find the temperature changes from 9080 to 2013 and this one is from 9080 to0 in terms ofra and again as you can see here this has been the changes from 90 80 203 and what exactly has been the original and then the rolling mean and the exponentially Ved mean and this one how exactly has been the rolling SD as well thank you so much for joining guys and have a great ahead take care bye-bye