Determination of the structure of a protein from the sequence of amino acids has been a major goal in computational biology and bioinformatics. A strong correlation between sequence entropy, and inverse packing density has been shown in recent studies indicated by the occurrence of two major regions, but with a lot of noise in the relationship data. One hundred and thirty query proteins and their sequence alignments are used to test modifications to sequence entropy calculations that significantly reduce the noise in the data. Gapped entropy, Gerstein-Altman entropy, and window average entropy offer improvement in terms of linear correlation but no significant improvement in the data noise is observed due to the introduction of 21st gap term, or Gerstein-Altman random entropy term. Averaging the sequence entropy that includes the 21st gap term within three neighbors resulted into smoothening of the entropy curve with no significant reduction in the data noise.