Stats Resource Project

The dataset summarises the price of used automobiles in Poland, based on the car brand, year of the car and its mileage. There are a total of eight (8) columns and 117’928 rows (first brown is the reader) in the dataset. These variables are, mark(car brand), model, year, mileage, volume of engine, fuel,province, and price.The mark represents the brand, where it was designed, manufactured from. In the dataset, we could see categories of Audi, Opel, Volvo, and others . Model, representing the vehicle model(ex. Volvo V90). The year represents the year it was manufactured. The mileage represents the total travel distance at the point of sale. This is recorded in kilometres. Volume of engine is how much fuel volume of fuel and air that can be pushed through a car's cylinders. This is measured in mL(cc). The variable fuel represents the type of fuels used for the vehicle. Here, in the dataset, it is able to observe gasoline, diesel, and others(some are electric, can’t classify as fuel that engine uses). The province shows which province it is listed/sold from within Poland. The price shows the price that the vehicle was sold at .There are 2 different categories of variables quantitative and categorical. The categorical variables include the mark (the car brand), type of fuel required, province in Poland and the model which are nominal, discrete variables. The quantitative variables include the year of manufacturing, mileage and price of the car. Mileage and price are ratio, continuous variables while the manufacturing date is an interval, discrete variable. Variable Names and Descriptions Variable Name Description Type of Variable mark Manufacture of the Vehicle (car brand); values are things like opel, audi, etc Categorical, nominal, discrete model Vehicle model; values are things like combo, astra, etc. Categorical, nominal, discrete mileage Distance on how much the vehicle travelled(kM) Quantitative, ratio, continuous vol_engine how much fuel volume of fuel and air that can be pushed through a car's cylinders, measured in litres Quantitative, ratio, continuous fuel Which type of fuel; values are things like diesel, gasoline, LPG, etc. Categorical, nominal, discrete province Which province the vehicles are listed/sold values are things like Lubelskie, Wielkopolskie, etc Categorical, nominal, discrete price Price of vehicle when sold, in Quantitative, ratio, continuous PLN (1 PLN=1 USD) Modifications to dataset: We removed two variables, the formatted generation name since it is not always existent as well as the city the car was sold in. Research question 1.​ Did the majority of cars made before 2017 sell for a price below 12’000 PLN? 2.​ What is the 65% estimate from mean caprice in 2017 in poland? 3.​ Is there a relationship between milage, and price? How does mileage affect the price of a used vehicle? 4.​ Is there any relationship with volume of engine with province, fuel type,and mark(some type of province tends towards a certain volume of engine due to landscape)? 5.​ If there is a higher mileage or older year of the car, how would the car price affect? Would it increase or decrease and how much? Aleksander Glotov(2022). Car Price Poland(version 1) [Data set]. Publisher Name.https://www.kaggle.com/aleksandrglotov/car-prices-poland Phase 2 Plan section The population of interest for this study was all vehicles in Poland as we desire to predict car prices in Poland. The sampling frame is the set of units the sample is drawn from, for this study the sampling frame was vehicles on sale in Poland(pending on its type and condition). The sample of the study is the 117,928 vehicles surveyed, their owners, for the purpose of the study. The sample was grouped into types of vehicles(brand,fuel type), and by its condition(mileage). The study design of the experiment is an observational study, various information is collected from the participants without manipulating the explanatory variable. The observational study was conducted as a survey to collect information about the study participants' vehicles. The owners would find all the information on their car by looking at all the details on their car and the price it would currently sell at. The sampling design used to collect the data was non-probability, voluntary response sampling. Individuals self-select to participate in this study by responding to an open invitation(anyone who volunteered was participating, this leads to bias). The invitation for the study was an open invitation on a Polish radio channel. The researchers don’t know who will receive the invite since it was undirected to the Polish population. The announcement on the radio invited car owners to complete a survey of various questions. We will plot the graph to see which variable(independent) affects the price(dependent). We will assume that with linear motion of the graph(or non-randomized plot), it would represent the relatedness of the variables. Data section From the phase 1 of the activity we had retrieved a data set of a variety of vehicles in Poland. By removing unwanted values from the chart, we were able to get condensed versions of data only with crucial data, important for graphing. On the chart we had, there were a total of 7 values. Mark, model, mileage, volume of engine, fuel, province, and price. By pairing each value with the price, we would consider which variables would affect the price change. If the variable we’re comparing with price is quantitative, it’ll be most appropriate to use a scatter plot. On the other hand with categorial, using boxplot(compare mean value) would be appropriate. We’re expecting to see a linear change of variable/price if both are related to each other(if the variable affects the price). For example , with the graph plot of mileage/price, as the mileage increases, the price would decrease in nearly linear style, it would be fair for us to assume both values are in relation to each other. The types of graphs would vary, by their components. For the numerical summary mean, the mean can summarise small and large samples of quantitative data. The numerical summary mean can be used for the research question If the car is 5 years old (2017 model) then what is the expected price of the car? The numerical summary called the five-number summary displays the minimum, first quartile, mean, third quartile and maximum value. It shows the measure of spread to characterise the variability of the distribution in the sample. A range can be calculated to determine the difference between the minimum and maximum. The graphical summary, scatterplot, can be used to help answer the research question: If the car is 5 years old (2017), what is the expected price of the car? Scatterplots are used to show a relationship between 2 quantitative variables. A.​ We can solve it by plotting a scatter plot. Since the variable ‘Year’ is quantitative. If the age of the vehicle and the price are in linear fashion, it would be ‘dependent value’ Additionally, if it’s linear, we’ll be able to notice the estimated price of a 5-year old vehicle. Examples on relationships This is a plot of price/volume of engine. It does not show linear fashion, therefore, it is hard to say two values are in relation to each other. On the other hand, this is a plot of price/mileage. Here we could clearly see there's a relationship present between two values, where larger the mileage, the price decreases(it’s common sense). Therefore, we could conclude price is dependent variable on mileage Here is the box plot of the fuel type consumed vs the price of these cars. From this figure, we see that CNP and LPG cars are the least expensive car types on average followed by diesel and gasoline. However, there is a significant number of cars that consume gasoline that are outliers and have a higher car price. Electric cars seem to be the most expensive car through the box plot. ​ Through graph plotting, we noticed there was a trend for the price of a car and based on its age(So it’s relevant). The data was retrieved in 2022, so the 5 year old vehicles would be manufactured in 2017. If we average the price of the car in 2017, we would get the estimated price of the vehicle(however, it is still the estimated average price of the 5-year old vehicle, the price may vary by its other independent variables). We can't really get any valuable information for the scatter plot above, so we’ll use the mean plot. Phase 3 One sample inference Hypothesis test of the proportion by a large sample test for proportion.By running hypothesis test, we would compare the mean of two(or more) groups. Question: Did the majority of cars made before 2017 sell for a price below 12’000 PLN? I'm going to make the hypothesis test on R Two sample inference By using confidence intervals, our group can find uncertainty of sample variables(around the mean). Question: What is the 65% estimate from mean caprice in 2017 in poland? Here, we would need a sample mean 2017 car price in Poland, we also would need standard deviation for it. Simple linear regression OR Multi-sample inference Multi-sample inference (ANOVA) is used to establish if statistical differences are present between mean of independent groups. Question: Group Reflection For the first part of the assignment, our group had gathered open resources from Kaggle, and sorted out data that wasn’t important for identifying our research question(for example in our raw data, factors such as owners, colour, series were not important). Next, to show the relatedness of the data visually, we have used rstudio to plot appropriate graphs for given variables. If only quantitative, used scatter, with both quantitative, categorial, and box plots. Also, plotting of R studio was also used to achieve quartiles, to have a brief understanding of the data we have before plotting(to notice any error if there's any). However, even though we’ve checked and removed any variable that could tackle from getting exact value/plot, some of our variables had shown multiple outliers, which was not understandable at first. However, since we were working with vehicles and their price, I’ve figured that some ‘collectible’ rare vehicles may impact these outliers for example, in the manufacture year/price graph there were unreasonably high prices compared to its manufacturing date. Through trials, we have left out outliers from the plot to have accurate estimation. If we had ignored any of those, we would’ve had a higher mean value than expected. this won’t be noticeable through quartile test since it only shows median which won’t impact that much( mean value also not shown in boxplot, so may be unnoticable without testing) I think there is a good chance, 80%, that I will refer to this Resource File in the future. In first semester, I took a course that used R and I know I would be apply to apply what I learned in this class and while completing this resource file in future courses and potentially even for a future job. I probably would come back to this material in 90% chance. As a future career, i probably would design and initiate many studies(individuals response on health effect/experience of medication). So I have a good chance of working with course materials in the future(rstudio, planning, managing data).