In this work we have analyzed the physics Concept Inventory (CI) scores and knowledge shortcomings in high school students, considering detrimental effects of the full-scale online learning due to the pandemic closure. This examination is performed through descriptive analysis based on the Force Concept Inventory approach and empirical modeling, by using the outcomes of dedicated tests which we conducted in various conditions and students' groups. Initially, we have assessed the CI scores and student's ability to answer the test by employing the Rasch technique, and next we have examined influential factors and their features, from a general methodical approach. In this regard, we have evidenced the non-decisiveness of mathematical (calculus) knowledge incapacities on physics knowledge failures and have employed this fact to confirm the conceptual knowledge failures as key concern linked with physics exam's unsatisfactory outcomes. It is used also to qualify the shortage on mathematics classes attendances as an independent factor in the modeling stage. Next, by considering the nature of the most frequent errors observed and dynamical properties of CI scores, the insufficiency on laboratory work, practices and demonstrations support for physics learning is identified as another important factor for conceptual knowledge failures observed. The teaching effectiveness issues, which herein is associated with communication technology performance and adaptiveness of teachers and students with online activities, has been investigated specifically, as another key factor. So, the procedural tests' faults are decomposed in conceptualization errors and calculus errors and all tests' shortcomings are classified corresponding to the resulting four error states. The estimation hereto is based on the Likert technique analysis. The findings of this step are interpreted in terms of the Force Concept Inventory analysis and six common misconceptions in Newtonian mechanics. Following those observations, this factor is recognized as an independent variable too, and for modeling purpose it has been embodied in a "location" variable based on the school's city ranking. Other expected influential factors, the attendance in physics and mathematics classes during online learning, the school type and category, and students' preference for physics, are analyzed simply in the stage of the modeling regression. Those factor variables used in linear and logistic model discussed hereto are taken categorical and their values have been assigned by simple conventions explained on the text, whereas their weightiness on the knowledge indicator variable is estimated directly from the regression. Next, using the predicted value for students' ability to solve a CI test, we proposed some variable configuration scenarios to allocate optimal strategies, which is defined herein as education activities that increase student's ability to solve a standard test, and keep difficulties unaffected or even improve them. We argued that within some extension, several findings of this work can be used more generally for similar system. [ABSTRACT FROM AUTHOR]