Coding: Be an ace of Python as well as R. There are different alternatives however these two are ubiquitous these days.
Know Thy Distributions: You ought to have a decent instinct of what appropriation is utilized for what. Given some data, you ought to have the option to accomplish something like this for some situations:
Fitting: When you have your distributions down, you should realize how to fit them to data in smooth manners. Start with greatest probability and go from that point.
Old style speculation testing: I think p-qualities and frequentist speculation testing when all is said in done are extremely difficult to disclose and difficult to comprehend (neglecting to dismiss invalid theories &c), however both are as yet ubiquitous.
Markov chains + ringers + whistles.
Fundamental Bayesian reasoning and demonstrating: Figure out how to consider everything a probability circulation rather than only a solitary worth (if proper). Have the option to amass the models and figure with them.
Some outdated details and probability hypothesis: For example “Irregular factors; transformations, restrictive desire, second creating capacities, assembly, limit hypotheses, assessment; Cramer-Rao lower bound, greatest probability assessment, adequacy, ancillarity, fulfillment. Rao-Blackwell hypothesis. Some choice hypothesis.”
Relapse! First direct, at that point non-straight. (Wheeze!)
Machine learning: I realize you said “statistics,” however on the off chance that you need to be a “data researcher” at that point machine learning will be an incredibly adaptable and valuable toolbelt for you. Likewise, machine learning is wide, so perhaps that could be another Quora question. =)
Composing: Convey your thoughts plainly, briefly, and compellingly.