<p>Cucumber (<i>Cucumis sativus</i> L.) though having high genetic diversity because of its origin in India yet systematic documentation of indigenous landraces is limited. The present study is aimed to determine the genetic control of major yield and disease resistance associated characters utilizing 29 indigenous genotypes and through artificial intelligence (AI) driven tool like Multiple linear regression (MLR) using generation mean analysis. The genotypes were grouped into eight distinct clusters based on most important characters governing yield and disease resistance. Of the 29 genotypic assemblage, 8 diverse genotypes were selected as parents to develop F₁ hybrids following 8 × 8 half diallel fashion without reciprocals. Multiple linear regression (MLR) revealed fruit weight, fruit length and number of fruits per plant as the major yield determining factors, while leaf blade length, total sugar content and ascorbic acid content influenced resistance to downy mildew. Based on manifestation of heterosis, 10 best performing hybrids were selected to develop 6 generations (P₁, P₂, F₁, F₂, BC₁ and BC₂). Generation mean analysis validated the predominance of dominance and epistatic interactions in controlling the most important characters. Overwhelming importance of non-additive gene action indicated the worth of implementing heterosis breeding. In the approach for developing high yielding and disease resistant line-bred variety, deferred selection for the concerned characters would be important. In this approach, AI-based crop modelling is combined with conventional breeding methods in an integrated framework for improvement of cucumber.</p>