Micro-fluidized beds (MFBs) with an ultra-fast energy transmission rate and high wall flux have recently attracted considerable interest. The hydrodynamic behavior in the MFBs has been demonstrated to deviate from the ones in the laboratory-scale fluidized beds (LFBs) because of the prominent wall effect. In order to understand the influence of the wall effect on flow regime transformation, a comprehensive experimental analysis, considering the effects of bed diameter, static bed height, and the properties of particles, was conducted using pressure drop data and visualization images. A new Hurst analysis, combined with a multi-scale resolution methodology, has been established to diagnose flow regimes, which successfully reflected the bubble characteristics of the fluidization system on the meso-scale. A generalized flow regime diagram was proposed based on the analysis of experimental data, and the influence of key factors on the velocity of flow pattern transformation was further investigated. On this basis, in the absence of preset function forms, the data-driven symbolic regression method was used to simultaneously search for the equation form and various parameters of the prediction correlation, and an empirical correlation formula for predicting the transformation of each flow pattern was automatically generated with excellent predictability. It is believed that this work is helpful for selecting desired fluidization conditions in practical applications, and this methodology can be expanded to the analysis of other complex systems with multi-scale characteristics.