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

it's a good morning good afternoon good evening guys based on the time zones you all are coming from so guys before we start with the session can you always give me a quick information if you all can see my screen and he will not increase well great thank you for confirming everyone so my name is neeraj keria and I have been working in the site industry for more than 13 years now before we proceed further let me quickly introduce our edu recap Master Class 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 back-end development Technologies and the best part about these webinars are they are absolutely free of course so there are no charges involved here and these seminars are agree are a really great platform for anyone who is looking to get into this Nursery 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 Hands-On 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 Wireless stationarity matters and environmental tools for the project including the Hands-On so if we talk about times three Zeta then what exactly it is it is a collection of observations of the obtained through repeated measurements over time and it also referred to as time Sam data is a sequence of data pointers indexed in time model so for example you can see here we have a data for a website visitors 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 you can see here we have the number of visitors per month and we have the timeline and then we have the average temperature recorded for the Yellowstone Park as a part of a Time series based data so what exactly a Time C's analysis is so if we talk about the entire analysis and time CS analysis is basically a statistical technique that deals with time series data or the trend analysis so the times three Theta means the data a 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 you 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 awesome we may have a question okay what exactly makes it special because see trans series is a collection of data pointers collected at constant time intervals and these are analyzed to determine the long-term trends so as to forecast the future or perform some other form of analysis but what exactly makes a tier is 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 Observer observations are independent doesn't hold in this case and along with an increasing or decreasing Trend most yes okay to the time series have some form of seasonality trends that is variation specific to a particular time frame for example we can see the sales of a Woolen jacket over time you will invariably find higher sales in winter seasons and because of the inherent properties of the 10 or the time series they are various steps involved in analyzing it and they are and they are just 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 see here we are talking about stock market data of 2020 from kaggle so kaggle is one of the primary resources 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 AV we can find data set for Aviation for banking for e-commerce for for the normal analysis based on 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 covered for economics for economics we have the country Regional and world GDP we have all happen to support 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 areas 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 treasury matters here so basically most of the times most time series models assume that each point is independent of one another and the statistical properties of data should not change over time stationary helps us better identify that if there are driving factors as well like we have the stationary time series and then we have a non-stationary Time series as the same data set as being used in these two different graphs and then we have environments and tools for the projects so basically here we do need to have the access to dump by pandas pad product daytime and SAS model so these are the libraries required here and we are going to work on Python programming language so now as a part of the Hands-On 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 pandas as PD super simple we are working on collab if you are comfortable with your own local jupyter notebook file then we can make use of jupyter notebook or we can make use of the other node profiles as per the requirement we can do that and for now we are going to make use of our 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 high-end system or we are experiencing a low performance in our own system then we can go ahead and make use of collab 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 PD then from daytime we are going to import date time and and time date then we are going to import mat.lib which is basically used for visualization part as plg then numpy as NP and then we have stats model to work on the statistical models here and then c bond which is basically an advanced visualization tool offer just like we have math problem and then we want to use inline so this line is being added here because we don't want to see the crafts being opened up 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 sometimes some warrants are going to be shown based on the data set and the method that we are going to make use of now we don't want that so we can simply go ahead and import one is all right so first of all before we start we have to load these libraries and then we have to we go ahead and write so for running the statement we can simply click on play and this is going to run the statement for us if we are going to work on uh we can say our 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 and suppose here we want to focus on GPU based instance in case we have we're going to work on multiple visualization tools then we can make use of GPU based kernel and again as you can see here now again the compute engine is currently being refreshed we can execute this again so all the major libraries required have been created all right so once we are done importing these then we are going to work on reading the CSV file if we have a global and temperatures by state DOT CSV so this is basically us data that we have the access to so in case you are going to work on collab so here we can simply put 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 Lan 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 is no need to correct 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 dot colab we have we are going to import drive so first of all we have to authenticate that yes we want to use a 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 is no need of mounting this in case you want to work on Google or we can say Cola drive then only we have to use it all right so here we can go ahead and so as you can see here we the drive has been mounted so now we can Define the path so our content is available under content 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 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 Lantern we just said by state dot CSV so this is a the file path that we have to Define all right so so we are going to make use of DF so the PD as in Panda so part 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 it doesn't that means the first five rows available in this feed and we are 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 a country all right so now if you want to see the types here we can send it on the DF type 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 sheet 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 currently going to represent the first five rows in terms of the first 100 rows being displayed returned as a response and now we are going to define the name so here we are going to rename the certain columns as D2 to date average temperature to average temperature average temperature on Saturday to do confidence intervalent uh conversional temperature to make sure that labels are more aligned to what we are trying to achieve here and then once we have changes we can see now the now this was the earlier heading right the other main column and again here we have the change heading and now we are going to make use of the same date to date time so here we are going to convert this column into daytime format itself and then we are going to set the date and then we are simply going to print the DF index so next we are going to Simply Define the different temperature changes so if we have we have latest differences in terms of the countries yeah Vision picture we had a group buyer by country and then we are going to Simply find the average and then we are going to sort the values based on average temperature as you can see the lowest has been for Canada that we have Russia U.S China Australia India and Brazil in terms of average temperature being defined and then we get simply plots on a graph as well where we have the figure size as mine four and then we are going to Simply import as 90 of four different values 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 readable again before that we can do that we have to ensure that we do run all the other data frames as well because again 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 are going to define the latest data frame as well suppose we want to create a temporary data frame from 1982 2013 and we can simply create a new data frame out of the existing data pin that we already have the access to all right and then we are going to create the latest latest difference so in terms of illustration we are going to define the country and the average temperature based on the new data pointers that we have included as average temperature so this isn't going to be a new view that has been defined and then we are going to create a new graph from VC later latest as you can see this is the latest brother has been created and then we are going to Simply work on resampling as well so after resampling we have simply resampled based on the parameters as a and then we assembly going to Simply redefine the resample parameters and then we are going to plot Theory sample uh plot by using the Matlock Library where we Define the title as the sperm figure size and then we are constantly plot temperature and here and we can see it has been continuously on a rise only it has been increasing order 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 series analysis so first of all we have to use a resample data frame and then from there we are going to exponentially by the find the weighted mean and then we are going to roll stand radiation and then we are going to create a supports next to each other and then we are simply going to to create two different graphs here temperature graphs we're rolling me and exponentially weighted mean as well and then we are going to create a temperature graph with rolling SRE where we're going to find the temperature change is from 1982 2013 and this one is from 1982 000 D in terms of the extra and again as you can see here this has been a changes from 1980 2003 and what exactly has been the original and then the rolling mean and the exponentially evaded mean and this one how exactly has been the rolling STD as well thank you so much for joining guys and have a great ahead take care bye