Cherry picking a data set for a single-variable analysis over a short period of time will simply not yield a meaningful conclusion in this kind of system. For example, if the latest warming trend was already underway before the industrial revolution, then disambiguating a human component vs. the "natural" chaotic component is impossible because we simply do not understand, in a predictable way, that chaotic system.
The very short data-set used to "prove" human impact on warming is not even a blink in the eye of time-scales involved in warming and cooling trends, and its magnitude is dwarfed changes from the chaotic system.