Title: Covid-19 data visualization, prediction and forecasting

Description: for this week's assignment we were given covid data and we had to use amazon sagemaker to visualize, predict and forecast. I had created a notebook instance called covid which is of type ml.t2.medium. I have directly uploaded the covid19 data in csv format in the notebook instance and created a file called model and set kernel to conda_python3. I have used matplotlib.pyplot and seaborn for data visualization, linearRegression and SVR for prediction and linear exponential smoothing i.e. holt model from time series analysis for forecasting. First of all I have loaded our dataset to the variable df and removed rows contains location as 'International' or 'World'. I wanted to visualize world data in accordance to days so I created another dataframe called datewise which groups the data in df in accordance to the date and agregates columns total_cases, new_cases, new_cases_smoothed, new_deaths, new_deaths_smoothed, new_tests, total_tests, new_tests_smoothed. I made barplot for new_cases, total_cases and total_deaths. I wanted to visualize data according to week so I added another column in datewise called "WeekofYear" and made two line graphs. One shows weekly pprogress of total cases and total death cases and another shows weekly progess of new_cases, new_death_cases and new_tests . Similarly I wanted visualized data accourding to location/county so I created another dataframe called countrywise which groups the data in df in accordance to the location and agregates columns total_cases, new_cases, new_cases_smoothed, new_deaths, new_deaths_smoothed, new_tests, total_tests, new_tests_smoothed and sorts them in ascending order. I also another column in countrywise called "mortality_rate" which hold mortality rate for each country. I create two sub plots which shows barplot of top 15 countries as per total_cases and as per total_deaths.In addition to above I create another two sub plots which shows barplot of top 15 countries as per new_cases and as per new_deaths. I also created barplot of mortality rate in accordance to the county/location. I also did data analysis for covid-19 for Nepal. I created barplots for total_cases, new_cases, total_deaths and new_deaths in Nepal according to date. For prediction and forecasting I wanted to input no of days since covid-19 outbreak and output the total cases in world at that particular day. For that I added another column in datewise called "Days Since" which contains no of days since covid19 outbreak. For prediction I seperated 95% of data for traing and 5% for validation and used linear regression and svm. For forecating I seperated 85% of data for training and 15% for testing and used holt model. I predicted / forcasted for 18 days after oct 12, 2020. The result of prediction for top 5 rows is below: Dates LR SVR Holts linear model predictions 0 2020-10-13 25970103 18156971 88847671 1 2020-10-14 26085091 18405054 89110309 2 2020-10-15 26200079 18656607 89372947 3 2020-10-16 26315068 18911666 89635585 4 2020-10-17 26430056 19170267 89898223