Properties of molecular clouds (MCs) lying in our Galaxy and their star formation scenarios have been investigated with the help of multivariate unsupervised machine learning techniques concerning several observable parameters. At first, the MCs have been classified into four coherent groups using the standard K-means clustering method. Subsequently, the optimum number of groups has been estimated by applying the Elbow method as well as the computation of Silhouette widths for a robustness check. Later, the properties of the groups are studied through several observable parameters as mentioned along with computed ones e.g. star formation rates (SFRs), virial masses, mass-spectra, dynamical time scales (Td), etc. to get a deeper understanding of the star formation process and dynamical evolution of these clouds. It is found that cluster 1 is suitable for the formation of field stars, binary pairs, or stellar associations, whereas the clouds in cluster 2 and cluster 3 are favorable sites for the formation of Galactic clusters of moderate masses, and cluster 4 may produce massive Galactic clusters as well as a few globular clusters. Surprisingly, for each cluster, clouds at around Galacto-centric radius ∼8 kpc, and on the near Galactic plane has a significantly low SFR. These occurrences indicate that the star formation phenomenon has yet not started or the proneness to start in that region.