The answer comes from the number of stars examined by each method.
Kepler in the first part of its mission examined 150,000 stars. After the extended mission it has examined 503,506 https://en.wikipedia.org/wiki/Kepler_space_telescope.
Kepler stared at one patch of sky at a time measuring the brightness of many 10's of thousands of stars at a time. The new TESS satellite mission will examine about 200,000 stars.
The radial velocity method has to take repeated measurements of each candidate star then move onto the next. For bright stars the exposure might be 2 minute, while fainter stars 10 minutes or so ... The HARPS instrument uses the majority of available nights on a 4m class telescope. The HARPS initial catalog of candidates was 376 https://phys.org/news/2011-09-exoplanets-harps.htmlstars. This has been expanded and changed over the years. There are several other major radial velocity searches. Their lists overlap so all-in-all they are checking 5,000 stars A personal estimate).
These 2 techniques are fairly comparable as they are both most sensitive to close in planets. So the difference in the number of plants found is because transit surveys examine more than enough stars to overcome the low probability of detecting a planet transiting it host star.
The other techniques for finding exoplanets favor finding different sorts of planets. For example microlensing tends to find high mass planets about the distance of Jupiter from its star. Of the millions of stars check for microlensing roughly 3,000 (now, far less in early searches) show microlensing in a year, of these only 10's have the signatures of planets. It typically takes a year or so to model each event ... so relatively few planets are found.
- Method number of stars examined number of planets found
- Transits 500,000+ 3126
- Radial velocity 5,000 778
- Microlensing 100's estimate 84
- Direct imaging 100 guess 47
- Astrometry 10 guess 1
the estimates of the number of stars are my estimates.