DMA is not the best pick. It is great for measuring Tg, and rubbery plateau, it may even work for some polymers, but not for those which soften up too much and too far before the melting point. It would be impossible to hold such samples in place when using standard DMA configurations. In that case, oscillatory rheology is probably the closest alternative based on the same principle but a different experimental setup adjusted to liquid samples, including melts. Dilatometry might also be an interesting solution. Nonetheless, measuring XRD at different temperatures is probably the most accurate and reliable method to measure crystal phase changes invoked by temperature.
Article High-temperature XRD investigations on phase transformations