More data in one place
Data is pulled from the source* on page load, then processed in real time to add important info that is annoyingly hard to get anywhere else. All the data that drives the graphs is displayed at the bottom of the page in sortable, filter-able form so you can dig into anything you notice in the graphs in excruciating detail. This includes the absolute numbers, i.e. not normalized or adjusted for population, as well as all of the errors and weirdness coming from the states.
Per capita data*
Removes the effect of population.
Five new cases in NY may be great news, in WY it may be ominous, because the populations are so different. As states take different approaches to quarantine, it is important to see the impact of those changes. The best way to make sense of events is to look at data adjusted for population. The charts below show data per 100k people.
Normalized data
A percentage of the state's own maximum.
At this point, the main thing we want to know is what's HAPPENING, not what happened. We want to know the trend - are things getting better or worse?
If a state has had a very high number of cases at some point, it distorts the graph, making it difficult to compare current trends with states that have had relatively few cases. Normalization divides the results by the maximum for that state. In this way we can compare, say NY with MN, and the trends are very clear regardless of the absolute number of cases.
Note: negative values are removed when calculating normalized and moving averages.
These values are erroneous attempts to compensate for errors in previous reporting, but have no meaning in daily values. Ideally, we could remove the erroneous spike, but that is harder to do than just ignoring them.
Moving Averages
Smooths out sharp peaks and valleys.
Data reporting is a hit and miss operation. A state may not report for a few days, then report an aggregate number for those days. It APPEARS that they had no cases, then they had a bunch, suggesting some special event occurred. It didn't! This results in very noisy data which makes comparing trends very difficult. The solution for this is to average results for a time period of several days and report that average. The longer the period, the smoother the graph. If we make it too long, then it starts to hide actual trends. An time period of 1 week was tried initially, because there is a 1 week noise pattern caused by weekends, but it was still noisy, so the period has been increased. The actual number of days used is displayed with each graph.
Note: While normalized data should have a max of 100%, time averaging reduces those peaks, so you will rarely see a metric reach 100% in these graphs.
Performance Sets
Winners and Losers
In addition to picking and comparing states as desired, you may load various Performance Sets (see Instructions below). These are the states that are winners, losers or doing OK for each metric (New Positive Tests %, New Cases, New Hospitalizations, etc.). By "Winner" we mean a state that is doing better, not more. E.g. A New Deaths Winner has decreasing deaths. These are selected based on the trend of the metric for the past week.
Instructions
(We've been adding features so fast, that these instructions are somewhat out of date. We hope to correct that soon. Thanks for your patience.)
If you have already visited and made state selections, those will be shown when you return. Otherwise, the charts start with NY, MN and ID visible - a high, medium and low incidence states. This will help make clear the meaning of each chart.
Display or hide any set of states for all charts using the Global State Selector.
- Click a state to select
- Shift-click states to select a group
- Command-click a state to add/remove it from selection
- Winners and Losers - you can click buttons to load states that are winners (doing better), losers (doing worse), or just doing OK for certain metrics. E.g. Losers for the "New Positives" metric. These do not update the your saved set (re-loading the page will restore those.)
- None, All - clear selection or show all states
- Performance sets: - loads set of states meeting the given criteria.
The set applies to all graphs. I.e. if you choose the states with increasing hospitalizations, Hosp Lose, those states will be shown in all graphs.
This one click approach along with no limitation of which or how many states you can show at a time enable you to easily analyze any scenario you desire. Compare all the red states. Compare all the blue states. Compare your state to your neighbors.