The first column of the CSV gives the Unix time at which the measurement was taken. The other two columns are the ADC output for respectively the TGS2602 and the TGS2600.
Using the full CSV data, we can now try to figure out when is the best time to trigger the Twitter alert. For now, it posts a new alert when one value varies by more than 30% compared to the last posted.
NB. The huge peaks are because of people playing with cigarettes near the sensors. Smoking kills.