Zhan Gong

---

• Background

My final project is based on a video game called Defense of the Ancients 2 (DOTA 2). As a sequel to Defense of the Ancients (DOTA), DOTA 2 is a multiplayer online battle arena game developed and published by Valve Corporation. The game has the highest prize pool in all kinds of the game championship. As of November 15, 2019, the total prize pool of Dota 2 International reaches $34,330,068, making it the most lucrative game in E-sport (Dota 2 Prize Pool Tracker 2019).

The game is played in matches between two teams of five players, with each player controlling one of the 117 heroes (each hero has his/her unique style and ability) and each team occupying and defending their own separate base on the map. During the match, players collect experience points and items for their heroes to successfully defeat the opposing team’s heroes, and there are 160 different kinds of items/pick-ups one can choose to defeat the opposing team’s heroes in Player versus Player combat.

By visualizing different heroes and their win rates, as well as the win rates of different items, this project seeks to provide players with guidance in selecting heroes and items, and to increase their chance of winning the game.

• The Research Questions

I ask four questions in this project: 1) Which hero has the best chance of winning the game? 2) Which item helps the player the most to win the game? 3) For each hero, does the win rate vary between different skills of the player (Normal / High / Very High / Tournament)? 4) For each item, does the win rate vary between different skills (levels) of the player?

• Audience

This final project means a lot to me. Before coming to the United States to pursue my Master’s study, I worked as a professor DOTA 2 player for four years.Therefore, I hope this project can benefit other professional DOTA 2 players like me and increase their competitiveness. Also, this project will be helpful for DOTA 2 beginners, who are not very skilled but willing to do better. Beginners will probably learn something from this project about how to choose heroes and items so that their likelihood of winning will be maximized.

• Data and Visualization

The data is obtained from a website called “DOTA Max.” The website collects data on every match that happened in the game. Variables include: the name of the hero (there are 117 heroes altogether); the win rate of each hero (0% – 100%); the name of the item  (there are 160 items altogether); the win rate of each item[1] (0% – 100%;); the skills of each player (Normal / High / Very High / Tournament) (DOTA MAX 2019)

I use Excel to extract the data from DOTA MAX. For every hero and item, I first collect the win rates of all players, and then divide the players into four subgroups according to their skills and collect the win rates of each subgroup players. I also provide information of each hero’s attribute (Agility/ Intelligence/ Strength), because this is a teamwork game, besides considering the win rate of the hero, finding a hero that fits the team’s demand is equally important.

I am using a bar chart for all these charts below. Because when I am using Tableau, I want to give the image of each hero, and the win rate is better for a bar chart to visualize. Also, I want this chart to be like a searching engine. To fix all the different player’s skills data in one dashboard and compare, I think a bar chart is the best choice.

Figure 1 Hero’s win rate

Figure 1 shows the win rate of each hero, and matches played during the second week of November.  Because DOTA 2 is frequently updating, this data is more accurate for the recent version. The heroes are showing on the left side, separated by attributes. Click on each hero to show the win rate and matches played by this hero.

Figure 2 Hero’s win rate by different skills

Figure 2 is showing the hero’s win rate by different skills of each player. Because of the difficulties of each hero, or higher skills player have some method in allusion to some heroes, heroes’ win rate are different.

Figure 3 Item’s win rate

The data of Item’s win rate based on which item does the player keeps until the end of the game. The visualization here is a commendation for the player choosing items during the game. Figure 3 is showing each item’s win rate and how many times they choose this item.

Figure 4 Item’s win rate by different skills

Figure 4 shows the item’s win rate by different skills of each player. Click on one of the items to show that item’s win rate and matches played by different skills.

References:

1. Dota 2 Prize Pool Tracker. 2019. “The International 2019 – Dota 2 Prize Pool Tracker.” Retrieved November 23, 2019 (https://dota2.prizetrac.kr/international2019).
2. DOTA MAX. 2019. “Heroes－ Dotamax – DOTA DOTA 2 Statistics Matches Live Dashboard.” Retrieved November 23, 2019 (http://www.dotamax.com/hero/rate/).

---

[1]  The win rate of an item is calculated at the end of the match, and only counts the items that the player still kept in the package.

• Research Questions

For this project, I am going to visualize my iPhone battery usage in the last 10 days. I just upgraded my phone to iPhone XS about a year ago. While this product is said to last for up to 12 hours of internet use – according to Apple’s advertised claims – I notice that I charge my iPhone more frequently than every 12 hours. This made me wonder, why is my phone’s battery draining so fast? Which apps contributed to my battery usage? What is the proportion of battery used for screen on activities and screen off activities, respectively? At what time of the day does my phone battery usage soar?

• Audience

I am the primary audience and beneficiary of this research. Despite that mobile phone constitutes an indispensable part of my life, I have never thought about how I use it every day and its impact on me. For example, a glimpse of my phone battery usage reveals that my screen was on for 6 hours yesterday, and I spent 10 hours in total on Candy Crush Saga in the past 10 days! Thus, understanding my phone battery data may serve as a reminder and help me spend my time wisely in the future.

Moreover, visualizing how long each app was in use on-screen or in the background allows me to make informed decisions to help prolong my phone battery life. For example, if an app is using a lot more battery power than others, I may remove or replace it. Or, if I know I need to save battery on my phone in certain periods of time, I can simply close out some apps or disable background operations within an app’s settings. In a word, this research will help me monitor and improve my life in the future.

• Data and Visualization

The data is gathered from my iPhone. iPhone collects battery usage data from both the last 24 hours and the last 10 days. Specifically, it includes the last charge level (0% – 100%), how long each app was in use (minutes) and its proportion of battery usage (%), battery usage during different time periods (e.g. 12am, 3pm, etc.), and the amount of time when the screen was on or the app was operating in the background (minutes).

I will visualize four things: 1) daily battery usage over the last 10 days; 2) the proportion of battery usage for the major apps over the last 10 days; 3) screen on activities and screen off activities over the last 10 days; 4) battery usage during different time periods in the last 24 hours (morning, afternoon, evening).

Figure 1 Daily Phone Activity in Minutes

Figure 1 shows the daily phone activities from Oct. 12^th to 21^st. According to the line chart, my phone has less activity on Oct. 19^th and 20^th, which are the weekends. On the contrary, the peak of my phone activities falls on the 15^th, which is a Tuesday. The fact that I rely more on my phone on weekdays may due to the fact that I was busying writing up my assignments and thus used Safari a lot (to search for references). I choose to use the line chart because I want to see the trend of my phone activities over the last 10 days.

Figure 2 Proportion of Battery Usage for the Major Apps from Oct.12^th to 21^st

Figure 2 talks about the proportion of battery usage for the major apps over the last 10 days. For example, I used Safari for 992 minutes during the last 10 days, which accounts for 35% of the total battery usage. Interestingly, the second most frequent app that I use is Candy Crush Saga, which accounts for 24% of the total battery usage – I really spent a lot of time on Candy Crush over last 10 days! Here I choose to use the pie chart, since it helps to visualize the proportion of each app’s battery usage.

Figure 3 Screen On Activities from Oct.12^th to 21^st

Figure 4 Screen Off Activities from Oct.12^th to 21^st

As I want to see which app uses a lot more battery power than others, I make two pie charts to compare the screen on activities and screen off activities. The sum of the screen on activities is 3,093 minutes, and the sum of the screen off activities is 742 minutes. Compare with Figure 2, the screen off time is not causing much battery usage. For example, Eudic (an English dictionary) uses 364 minutes of screen off time but just accounts for 1 percent of the total battery usage. However, even though I haven’t opened QQ Mails, it still causes 1 percent of battery usage. Therefore, a wiser way is to turn off all the apps while not using it.

Figure 5 Phone Activity during Different Time Periods in the Last 24 Hours

In Figure 5, I visualize the phone activity during different time periods in the last 24 hours. I choose line chart again, because it can visualize the trend in data over intervals of time. As we can see, I use my phone a lot in the morning and at night. Except for nighttime while I am sleeping, I use my phone almost all the time. It should be noted that, however, the data includes the off screen activities as well, therefore, it might not be accurate enough.

• Next step

The data only includes 10 days of the phone activities; any data longer than 10 days will be erased. Future study should collect data over a longer period of time, for example, a month, or a year, to make the conclusion more convincing. In addition, I would like to calculate each app’s battery usage per minute to further investigate which app is more energy efficient.

• Research questions

In 2019 (from January 1^st to September 24th), New Yorkers made 1,869,791 requests for service via 311. Among them, illegal parking accounts for 7.6% of all requests (n=141,444), making it one of the major complaint types. For this project, I am interested in all complaints made in 2019 related to illegal parking. I ask, 1) What do people usually complain about illegal parking (For example, blocked bike line/blocked hydrant/blocked sidewalk/commercial overnight parking/posted parking sign violation, etc.)? 2)Which borough has the most illegal parking complaints, and why? 3) Are there any connections between illegal parking complaints and different seasons?

• Audience

My interest in this project stemmed from my personal experience. It was my first year living in New York, once my friend gave me a ride to Queens. On our way, she needed to stop by a post office, but we couldn’t find any parking area close to the post office, nor did we see any “no parking” signs. We decided to take a risk and parked the car to the side of the road temporarily, which resulted in a parking violation ticket when we returned. This made me realize the importance of understanding what behaviors were considered as “illegal parking.” By visualizing different types of illegal parking complaints and their frequencies respectively, New Yorkers may be more aware of how to “park legally” in the future.

   Understanding illegal parking complaints could also help city planners and policymakers measure neighborhood conditions and make better policies. For example, if certain neighborhoods/boroughs have much more illegal parking complaints than others, it could be a sign that more parking enforcement and government intervention are needed in this area. It is also possible that other factors, such as different seasons, may impact the volume of complaint calls. Perhaps people complain more over the summer. If so, developing a specific parking rule for the summer may help reduce the volume of complaints. Thus, it is worth exploring the fluctuations in the number of complaints over different months and understands its logic.

• Data and Visualization

The data is gathered from 311 Service Requests in NYC OpenData. I narrow my project to one complaint type, which is illegal parking, and I only look at the year 2019. In other words, the data includes all illegal parking complaints made in all five boroughs of New York City, organized by date from January 1^st, 2019 to present. The data also includes the date of the call, the borough of the incident and its geospatial data, the complaint type, the responding agency, and the action taken on a given complaint.

I will visualize three figures to answer my research questions: 1. Illegal parking complaint by borough, 2019; 2. Types of illegal parking complaint, 2019; 3. Illegal parking complaint by month, 2019.

• Visualization

First, I want to visualize and compare which borough has the most illegal parking complaints. As we can see from Figure 1, Brooklyn has the most illegal parking complaints in 2019, followed by Queens, Manhattan, Bronx, while Staten Island has the least illegal parking complaints in 2019. I choose the bar chart because there are just five boroughs (i.e. they are discrete categories of data), and a bar chart is easy to display and compare the frequency of complaints made by different boroughs.

Next, I visualize different types of illegal parking complaints. From Figure 2, it seems that people complain about posted parking sign violations and blocked hydrants the most. The fact that people complain a lot about blocked hydrant reflects individuals’ willingness to intervene in the interest of community well-being and their concern for the city. Again, since I want to know what people usually complain about illegal parking (frequency), and different types of illegal parking are discrete categories, thus, I use the bar chart to compare.

Finally, I visualize the illegal parking complaints that happened in each month. I choose the line chart here because I want to compare the number of complaints in different months and seasons, and line charts can best visualize the trend in data over intervals of time. According to Figure 3, we can see most complaints happened in May, June, July, and August, with July having the most illegal parking complaints, which leads us to make the conclusion that during the summer, there are more illegal parking complaints. Probably due to the summer heat, people are less patient and tolerant towards illegal parking in the summer; or perhaps there are more people who are outdoors and park illegally in the summer than in other seasons.

• Next steps

We already know that Brooklyn has the most illegal parking complaints, while Staten Island has the least illegal parking complaints. Further research should look into different neighborhoods and see if there are any variations in the volume of complaints within each borough, and if there are certain patterns between different illegal parking types and certain neighborhoods. For example, maybe one specific neighborhood has most complaints related to the blocked bike lane, while another neighborhood has most complaints related to commercial overnight parking. These patterns will allow us to further explain and support our preliminary findings, and lead policymakers to make more targeted policy changes.

Welcome to CUNY Academic Commons. This is your first post. Edit or delete it, then start blogging!