Project 2 – Fata Police Shootings – Revised Submission
Dec 1
In today’s analysis of economic indicators, I tried to visually explore the time series of crucial variables in our dataset. I used the seaborn library, by plotting a line plot that showcases the evolution of key factors over time. The variables i used include logan_passengers, total_jobs, unemp_rate, med_housing_price, and housing_sales_vol. The resultt is a time-traveling journey across dates, where each line on the plot explains these aspects of variables. It’s like creating a historical timeline of the given dataset that allowed me to grasp the patterns, trends, and potential connections between these variables.
After that for Time Sereis Analysis, i’ve utilized the seasonal_decompose function from the statsmodels.tsa module to break down time series data into its key components: trend, seasonal, and residual. Specifically, I’ve chosen to decompose three variables—total jobs, unemployment rate, and median housing price—from my dataset. The seasonal_decompose function helps me understand the underlying patterns and fluctuations in these variables over time.
For each chosen variable, I’ve created a set of subplots using matplotlib, displaying the observed values, trend, seasonal component, and residual component. The observed plot shows the raw data, while the trend plot reveals the long-term direction, the seasonal plot highlights recurring patterns, and the residual plot captures what’s left unexplained. This visual breakdown helped me gain insights into the individual elements that contribute to the overall behavior of each variable.
After that, for insights into the overall direction and behavior of ‘total_jobs’ over the forecasted period, I plotted a linear trend that signifies a consistent, straight-line pattern in the variable’s values.
For that, I performed time series forecasting using the ARIMA (AutoRegressive Integrated Moving Average) model for the ‘total_jobs’ variable in the dataset. I began by importing the functions from the statsmodels.tsa module,; seasonal_decompose for decomposing the time series and ARIMA for building the forecasting model. The time series data for ‘total_jobs’ was extracted for the period from January 2013 to December 2019, which served as the training data for the model.
I then iinitiated an ARIMA model with a specified order of (1, 1, 1) . The model was then fitted to the training data using the fit() method. After fitting, I forecasted future values for the ‘total_jobs’ variable, specifying the number of steps to forecast into the future (in this case, 12 steps or months).
The visual output of the code is a graph that showcases the actual time series values for ‘total_jobs’ as a blue line and the forecasted values as a red line. The x-axis represents time (date), and the y-axis represents the values of the ‘total_jobs’ variable. The linear progressive graph indicates a trend in the ‘total_jobs’ data, suggesting a systematic increase or decrease over time. In this context, .
Finally, I’ve crafted a web app that acts as a control panel for delving into the depths of our dataset. Picture the dataset as a treasure chest full of various gems, each gem representing different aspects like jobs, housing prices, and more. The app zooms in on two particular gems: total jobs and median housing prices. The dropdown menus act like magical selectors (X and Y) that allow me to choose which gems to compare. With the first dropdown (X), I can pick what I want on the left side of the comparison (let’s call it the x-axis), say the total number of jobs. The second dropdown (Y) lets me choose what I want on the right side (the y-axis), perhaps the median housing prices.
Now, imagine the scatter plot in the middle as a mystical map that shows where these selected gems interact. Initially, the map shows something (we’re not sure what) just to fill the space. As I change my choices with the dropdowns, the map updates, unveiling how total jobs relate to median housing prices. It’s like having a window into the dataset, allowing us to witness patterns and connections between these two key aspects.
This setup essentially turns the exploration of data into an interactive adventure. It’s a dashboard where I pick two things, and it throws a visual representation to help me understand the correlation bw those two things.
Think of X and Y as two aspects we want to compare.
It’s basically like setting up a control panel for exploring the data. The two drop-down menus allows us to pick X and Y from our data.