CURRENTLY, SEVERAL COUNTRIES ARE SEEKING TO CHANGE THEIR ENERGY MATRICES TOWARDS MORE SUSTAINABLE SOURCES. IN CHILE, ONE OF THE RENEWABLE SOURCES WITH INCREASED PARTICIPATION IS PHOTOVOLTAICS. HOWEVER, PHOTOVOLTAIC ENERGY SOURCES HAVE AN INTRINSIC VARIABILITY, WHICH COMBINED WITH VARIABLE DEMAND IMPOSES A CHALLENGE FOR PROPER DESIGN. CURRENTLY, TOOLS AVAILABLE FOR THE STUDY OF THIS VARIABILITY ARE EITHER COMPLEX OR EXPENSIVE. WITH THE ADVENT OF DIGITALIZATION, THERE IS AN OPPORTUNITY TO INCORPORATE TOOLS BASED ON ARTIFICIAL INTELLIGENCE TO IMPROVE FORECASTING FOR MEDIUM AND LOW POWER INSTALLATIONS. THIS WORK PRESENTS AN APPLICATION OF MACHINE LEARNING TOOLS FOR IRRADIANCE AND POWER CONSUMPTION FORECASTING. THE METHODOLOGY IS INTENDED TO BE IMPLEMENTED AS A LOW COST SOLUTION FOR SMALL SCALE GENERATION. THE RESULTS SHOW THAT IT IS POSSIBLE TO PREDICT IRRADIANCE AND ENERGY CONSUMPTION THROUGH HISTORICAL DATA, CONCLUDING THAT THE METHODOLOGY BASED ON MACHINE LEARNING IS ABLE TO SUPPORT THE DECISION MAKING FOR THE IMPROVEMENT OF PHOTOVOLTAIC SYSTEMS.