This page describes the methodologies used to gather the data of City Filter.
Before diving in, I would like to make it clear that while numbers can give a rough indication of what a city is typically like, they could be very different from your experience there. With very bad luck you can get mugged in the safest city, have rain non-stop in the sunniest city, or experience an earthquake in a normally earthquake-free zone. Even the data that I tag as "Reliable" can be very different from your experience there depending on your luck.
The cost per day is calculated by adding the cost of living (source: Numbeo) per person and the typical average Airbnb cost per night for an entire place or hotel room. It is district-based, so make sure you expand a city to see the districts details. I use a median Airbnb price to exclude very expensive or very cheap places.
This crowdsourced data from Numbeo represents the perceived crime at the city-level (or country-level as a fallback). Please also note that criminality can greatly vary from district to district.
This is the Social Progress Index tweaked to range from 0 to 100. This index takes about 60 parameters into consideration, such as nutrition and medical care, access to knowledge, or personal freedom, and should represent pretty well how "developped" a country is (although that's a very debatable topic). If the SPI is not available for a country, I fallback to the average SPIs of countries that have a similar Human Development Index.
This crowdsourced data from Numbeo represents the perceived quality of health care services and the patients satisfaction at the country-level.
This comes from the Wikipedia page of a city. It is the daily mean temperature, or the average of the highs and the lows of a given month if the daily mean temperature is not available. The color scale is my personal interpretation of the felt temperature. To give you an idea, the flashy green color (implicitly meaning the "best" temperature in my opinion), is the lowest temperature where I can wear a t-shirt, shorts, and flip-flops at night without feeling cold (with that kind of clothes I feel cold under 20°C / 68°F). This may be too hot for your taste during the day, but that kind of gives a point of reference.
Anyway, look at a city you know well and compare it to others. If pale green or light orange is your favorite temperature for instance, just aim for these colors.
Data from Wikipedia. Note that I use white dots to represent snowy days when the average temperature is below 0°C, but those months can have regular rain, not just snow.
The values from GEM are the Peak Ground Acceleration (multiplied by 100), which represent the seismic hazard risk. When gathering that data for a city, I actually use the highest risk point in a 100km radius from the city, since earthquakes can cause damage far from their epicenter.
This data comes from... me looking at maps, and considering where earthquakes could happen and cause tsunamis. It is very much not accurate at all, but gives a rough idea of the risk of a district. It is district-based and not city-based. If I have doubts, I choose the higher risk value. Safety first.
This also comes from me looking at maps of all recorded cyclones trajectories, and from typical cyclones frequency per month. Cyclones (or hurricanes, typhoons, or tripocal storms), can happen any time of the year, and their trajectory may be atypical, so my estimates can be very inaccurate.
The population data comes from the Wikipedia page of a city. I take the Municipality or Urban area value (whichever is available), not the Metropolitan area value.
The pollution data is the Air Quality Index (AQI) from the World Quality Index for PM2.5 particles. This data may be inaccurate because of malfunctions of measurement stations, can vary from district to district, and is unavailable for many cities.
This is the number of active cases of Coronavirus / COVID-19 by country from Wikipedia as of March 13th, 2020. "Active" meaning that the number of death and recoveries is subtracted from the confirmed cases.