COMBINED COMPTON-SCATTERING AND PAIR-CREATION TELESCOPES ARE A KEY TECHNOLOGY FOR FUTURE OBSERVATIONS IN THE GAMMA-RAY ENERGY RANGE. THEY ARE THE DESIGN OF CHOICE OF TWO ENVISIONED FUTURE SATELLITE MISSIONS COMPAIR/AMEGO (MOISEEV+ 2017) AND E- ASTROGAM (DE ANGELIS+ 2017) AND WILL ALLOW US TO PERFORM UNPRECEDENTED MEASUREMENTS OF EXTREME PARTICLE ACCELERATORS (E.G. GRBS AGN) STUDY THE LIFE CYCLE OF MATTER GENERATED IN STARS AND SUPERNOVAE SEARCH FOR DARK MATTER SIGNATURES AND MUCH MORE. THE STANDARD ANALYSIS TOOLSET USED BY BOTH COMPAIR/AMEGO AND EASTROGAM IS MEGALIB (E.G. ZOGLAUER+ 2006). ITS DEFAULT SET OF ALGORITHMS GOES BACK TO THE DEVELOPMENT OF THE MEGA COMPTON AND PAIR TELESCOPE (ZOGLAUER 2005). HOWEVER WHILE THE EVENT RECONSTRUCTION FOR UNTRACKED COMPTON EVENTS HAS BEEN ADVANCED IN THE CONTEXT OF COSI (E.G. KIERANS+ 2017) THE INTERMEDIATE ENERGY BAND (2-3 MEV TO ~20 MEV) RANGING FROM TRACKED COMPTON EVENTS TO LOW-ENERGY PAIR EVENTS HAS NOT SEEN SIMILAR ADVANCES. THIS ENERGY RANGE FACES SEVERAL KEY DATA SCIENCE CHALLENGES WHICH HAVE TO BE OVERCOME SUCH AS (1) DETERMINING THE DIRECTION-OF-MOTION OF SHORT ELECTRON TRACKS WITH 2-4 INTERACTING LAYERS (2) DISTINGUISHING SINGLE TRACKS FROM COMPTON RECOIL ELECTRONS WHICH PERFORM TURNAROUNDS (E.G. GOING FROM UPWARD TO DOWNWARD MOVING) FROM LOW-ENERGY PAIR EVENTS WITH LARGE OPENING ANGLES (3) DECIDING IF THE EVENT IS A CHARGED PARTICLE SHOWER PAIR OR COMPTON EVENT WITH A SINGLE OR MULTIPLE TRACKS (4) DETERMINING ENERGY LOST IN PASSIVE MATERIAL FOR ELECTRONS GOING FROM TRACKER TO CALORIMETER OR IN THE STRUCTURE OF THE TRACKER AND IN NON-FUNCTIONING STRIPS/WAFERS (BOTH ARE ESPECIALLY IMPORTANT FOR COMPTON EVENTS WHICH REQUIRE THAT THE FULL ENERGY IS MEASURED FOR CORRECT RECONSTRUCTION) (5) DISTINGUISHING THE HITS IN THE CALORIMETER CAUSED BY ELECTRONS MOVING FROM TRACKER TO CALORIMETER FROM COMPTON INTERACTIONS OF THE SCATTERED GAMMA RAY (6) DETERMINING THE FULL ENERGY OF ELECTRON AND POSITRON FROM SOMETIMES VERY NEARBY INTERACTIONS IN THE CALORIMETER AS AN UNEVEN ENERGY DISTRIBUTION BETWEEN ELECTRON AND POSITRON INFLUENCE THE ORIGIN DETERMINATION AND (7) IDENTIFYING BREMSSTRAHLUNG HITS IN THE DETECTOR ESPECIALLY FOR COMPTON EVENTS AS THEY NEED TO BE DISTINGUISHED FROM ADDITIONAL COMPTON SCATTERS. FOR COSI WE HAVE RECENTLY SEEN SIGNIFICANT IMPROVEMENTS IN THE PERFORMANCE OF THE EVENT RECONSTRUCTION APPROACHES FOR (UNTRACKED) COMPTON EVENTS USING SUPERVISED MACHINE-LEARNING APPROACHES SUCH AS RANDOM FORESTS OF BOOSTED DECISION TREES AND NEURAL NETWORKS TO FIND THE CORRECT PATH OF THE GAMMA RAYS IN THE COSI DETECTORS. HERE WE PROPOSE TO APPLY THE LATEST MACHINE LEARNING APPROACHES TO THE RECONSTRUCTION OF TRACKED COMPTON EVENTS AND LOW-ENERGY PAIR EVENTS. THE APPROACH TAKEN WILL BE SIMILAR TO WHAT WAS DONE FOR COSI: USING SIMULATIONS WHERE THE EVENT TYPE AND PATH ARE WELL KNOWN TRAIN A SET OF DIFFERENT MACHINE LEARNING APPROACHES (RANDOM FORESTS VARIOUS NEURAL NETWORKS) TO PERFORM: (1) DECISION OF EVENT TYPE (2) CLUSTERING OF HITS WHICH BELONG TOGETHER E.G. TRACKS (3) DETERMINING START POINT OF TRACKS (4) DETERMINATION IF HITS ARE FROM COMPTON INTERACTIONS OR BREMSSTRAHLUNG (5) ESTIMATE ENERGY LOSSES IN IN PASSIVE MATERIAL (6) DETERMINING THE COMPTON PATH USING ELECTRON TRACK INFORMATION AND MORE. THE BEST METRIC TO DETERMINE THE PERFORMANCE OF THE NEW APPROACHES IS THE SAME AS THE ONE USED FOR COSI THE ANGULAR RESOLUTION MEASURE ARM AS WELL AS THE SCATTER PLANE DEVIATION FOR THE TRACK DIRECTION. ALL CODE DEVELOPMENTS WILL BE AVAILABLE WITHIN THE PUBLICLY AVAILABLE MEGALIB FRAMEWORK. THIS PROPOSAL FOR "DEVELOPMENT OF NEW DATA ANALYSIS METHODS FOR FUTURE SATELLITE MISSIONS" WILL HELP UNLOCK THE FULL PERFORMANCE OF FUTURE TRACKING COMPTON AND PAIR TELESCOPES IN THE DIFFICULT TRANSITION REGION AND ULTIMATELY FURTHER NASA'S GOAL TO "DISCOVER HOW THE UNIVERSE WORKS".