We describe the spline histogram algorithm which is useful for visualization of the probability density function setting up a statistical hypothesis for a test. The spline histogram is constructed from discrete data measurements using tensioned cubic spline interpolation of the cumulative distribution function which is then differentiated and smoothed using the Savitzky-Golay filter. The optimal width of the filter is determined by minimization of the Integrated Square Error function. The current distribution of the TCSplin algorithm written in f77 with IDL and Gnuplot visualization scripts is available from http://www.virac.lv/en/soft.html
翻译:我们描述用于为测试设定统计假设假设的概率密度函数可视化的 Spline 直方图算法。 Spline 直方图是根据离散数据测量构建的,使用Savitzky-Golay 过滤器对累积分布函数进行张力立方柱内插,然后对累积分布函数进行区分和平滑。过滤器的最佳宽度由最小化方块错误函数确定。以 IDL 和 Gnuplot 可视化脚本以 f77 撰写的 TCSplin 算法当前分布情况见http://www.virac.lv/en/oft.html。