Users' Daily Activities Data Visualisation
We use python to analyse each affective factor based on long-term data logging. Specifically, the more data we get, the more accurate the analysis will be. In this project. After calculating people’s influence on various factors (including circadian, nap, caffeine, alcohol, and meal). We want to visualise this data on the prototype. By doing this, people could view which factors affect them more and which are less, so that users could view the analysis outcomes more directly. The first figure displays the radar char, and the second is generated by JavaScript.
Technically we used JavaScript Library chart.js to visualise these data. We could see that the chart above that. The data we collected from this user indicates that the dominating factor that affects its sleep quality is circadian, which means probably frequently changed circadian clock (such as always stay up) is the main reason while other factors affect less. Additionally, as users use software for longer periods, the system can already find different characteristics for each user based on the data. (the chart below records the analysis outcome of 5 days, 15 days, and 30 days). Record data could be stored in a CSV file and specific data format (including how many days this app has recorded, and 5 main affective factors).
JavaScript reads a set of data from a CSV file each time by using fetch(“URL”) function, which means the analysis chars will be generated automatically when data is accumulating.
Then the outcomes from Chart.js could be used in the interface of analyses.
From the radar chart, we could see the distribution of how each factor affects the user’s sleep quality. More specifically, we could conclude that circadian is the most dominating factor that affects this user’s sleep. And other factors didn’t affect too much because of their low sensitivity coefficients. Therefore, when we design the suggestion function, we could focus more on the circadian factor, such as avoiding too much stay up, and how to form healthy and regular living habits.
If we iteratively analyse this function in future designs, we may consider how each factor affects explicitly the results calculated by the system. But, so far, we have not found much literature on the value of each specific factor (such as coffee) for users with differences. So in the next design, we may need to pay attention to the living habits of certain groups for a long time. Target users are more accurately divided, and different calculation methods and analysis results are tailored for them.